Fri, 11 Jan 2019 19:28:40 +0200
Donwloading of new products (Hirelpp, cloudmask, etc.)
victor@7 | 1 | import requests |
moritz@29 | 2 | |
ioannis@17 | 3 | try: |
ioannis@17 | 4 | import urllib.parse as urlparse # Python 3 |
ioannis@17 | 5 | except ImportError: |
ioannis@17 | 6 | import urlparse # Python 2 |
victor@7 | 7 | |
victor@7 | 8 | import argparse |
madrouin@24 | 9 | import datetime |
madrouin@24 | 10 | import logging |
victor@7 | 11 | import os |
victor@7 | 12 | import re |
ioannis@32 | 13 | from io import BytesIO |
madrouin@24 | 14 | import sys |
victor@7 | 15 | import time |
madrouin@24 | 16 | import urlparse |
victor@7 | 17 | from zipfile import ZipFile |
madrouin@24 | 18 | |
i@14 | 19 | import yaml |
victor@7 | 20 | |
madrouin@24 | 21 | requests.packages.urllib3.disable_warnings() |
i@14 | 22 | logger = logging.getLogger(__name__) |
victor@7 | 23 | |
victor@7 | 24 | # The regex to find the measurement id from the measurement page |
victor@7 | 25 | # This should be read from the uploaded file, but would require an extra NetCDF module. |
victor@7 | 26 | regex = "<h3>Measurement (?P<measurement_id>.{12}) <small>" |
victor@7 | 27 | |
victor@7 | 28 | |
victor@7 | 29 | class SCC: |
madrouin@24 | 30 | """A simple class that will attempt to upload a file on the SCC server. |
i@14 | 31 | |
victor@7 | 32 | The uploading is done by simulating a normal browser session. In the current |
madrouin@20 | 33 | version no check is performed, and no feedback is given if the upload |
madrouin@20 | 34 | was successful. If everything is setup correctly, it will work. |
victor@7 | 35 | """ |
victor@7 | 36 | |
i@14 | 37 | def __init__(self, auth, output_dir, base_url): |
moritz@29 | 38 | |
victor@7 | 39 | self.auth = auth |
victor@7 | 40 | self.output_dir = output_dir |
i@14 | 41 | self.base_url = base_url |
victor@7 | 42 | self.session = requests.Session() |
moritz@29 | 43 | self.session.auth = auth |
moritz@29 | 44 | self.session.verify = False |
i@31 | 45 | |
i@14 | 46 | self.login_url = urlparse.urljoin(self.base_url, 'accounts/login/') |
i@31 | 47 | self.logout_url = urlparse.urljoin(self.base_url, 'accounts/logout/') |
i@31 | 48 | self.list_measurements_url = urlparse.urljoin(self.base_url, 'data_processing/measurements/') |
i@31 | 49 | |
i@14 | 50 | self.upload_url = urlparse.urljoin(self.base_url, 'data_processing/measurements/quick/') |
ioannis@34 | 51 | self.download_hirelpp_pattern = urlparse.urljoin(self.base_url, |
ioannis@34 | 52 | 'data_processing/measurements/{0}/download-hirelpp/') |
ioannis@34 | 53 | self.download_cloudmask_pattern = urlparse.urljoin(self.base_url, |
ioannis@34 | 54 | 'data_processing/measurements/{0}/download-cloudmask/') |
ioannis@34 | 55 | |
moritz@29 | 56 | self.download_preprocessed_pattern = urlparse.urljoin(self.base_url, |
moritz@29 | 57 | 'data_processing/measurements/{0}/download-preprocessed/') |
moritz@29 | 58 | self.download_optical_pattern = urlparse.urljoin(self.base_url, |
moritz@29 | 59 | 'data_processing/measurements/{0}/download-optical/') |
moritz@29 | 60 | self.download_graph_pattern = urlparse.urljoin(self.base_url, |
moritz@29 | 61 | 'data_processing/measurements/{0}/download-plots/') |
ioannis@34 | 62 | self.download_elic_pattern = urlparse.urljoin(self.base_url, |
ioannis@34 | 63 | 'data_processing/measurements/{0}/download-elic/') |
i@14 | 64 | self.delete_measurement_pattern = urlparse.urljoin(self.base_url, 'admin/database/measurements/{0}/delete/') |
i@31 | 65 | |
i@14 | 66 | self.api_base_url = urlparse.urljoin(self.base_url, 'api/v1/') |
i@31 | 67 | self.api_measurement_pattern = urlparse.urljoin(self.api_base_url, 'measurements/{0}/') |
i@31 | 68 | self.api_measurements_url = urlparse.urljoin(self.api_base_url, 'measurements') |
i@14 | 69 | |
i@14 | 70 | def login(self, credentials): |
i@31 | 71 | """ Login to SCC. """ |
victor@7 | 72 | logger.debug("Attempting to login to SCC, username %s." % credentials[0]) |
moritz@29 | 73 | login_credentials = {'username': credentials[0], |
moritz@29 | 74 | 'password': credentials[1]} |
victor@7 | 75 | |
i@14 | 76 | logger.debug("Accessing login page at %s." % self.login_url) |
victor@7 | 77 | |
victor@7 | 78 | # Get upload form |
moritz@29 | 79 | login_page = self.session.get(self.login_url) |
victor@7 | 80 | |
i@31 | 81 | if not login_page.ok: |
i@31 | 82 | raise self.PageNotAccessibleError('Could not access login pages. Status code %s' % login_page.status_code) |
victor@7 | 83 | |
i@31 | 84 | logger.debug("Submitting credentials.") |
victor@7 | 85 | # Submit the login data |
i@14 | 86 | login_submit = self.session.post(self.login_url, |
moritz@29 | 87 | data=login_credentials, |
victor@7 | 88 | headers={'X-CSRFToken': login_page.cookies['csrftoken'], |
moritz@29 | 89 | 'referer': self.login_url}) |
victor@7 | 90 | return login_submit |
victor@7 | 91 | |
victor@7 | 92 | def logout(self): |
i@31 | 93 | """ Logout from SCC """ |
i@31 | 94 | return self.session.get(self.logout_url, stream=True) |
victor@7 | 95 | |
ioannis@32 | 96 | def upload_file(self, filename, system_id, rs_filename=None, ov_filename=None, lr_filename=None): |
madrouin@20 | 97 | """ Upload a filename for processing with a specific system. If the |
victor@7 | 98 | upload is successful, it returns the measurement id. """ |
victor@7 | 99 | # Get submit page |
moritz@29 | 100 | upload_page = self.session.get(self.upload_url) |
victor@7 | 101 | |
victor@7 | 102 | # Submit the data |
victor@7 | 103 | upload_data = {'system': system_id} |
victor@7 | 104 | files = {'data': open(filename, 'rb')} |
victor@7 | 105 | |
madrouin@20 | 106 | if rs_filename is not None: |
i@31 | 107 | logger.debug('Adding sounding file %s' % rs_filename) |
madrouin@20 | 108 | files['sounding_file'] = open(rs_filename, 'rb') |
madrouin@20 | 109 | |
ioannis@32 | 110 | if ov_filename is not None: |
ioannis@32 | 111 | logger.debug('Adding overlap file %s' % ov_filename) |
ioannis@32 | 112 | files['overlap_file'] = open(ov_filename, 'rb') |
i@31 | 113 | |
i@31 | 114 | if lr_filename is not None: |
i@31 | 115 | logger.debug('Adding lidar ratio file %s' % lr_filename) |
i@31 | 116 | files['lidar_ratio_file'] = open(lr_filename, 'rb') |
i@31 | 117 | |
ioannis@32 | 118 | logger.info("Uploading of file(s) %s started." % filename) |
victor@7 | 119 | |
i@14 | 120 | upload_submit = self.session.post(self.upload_url, |
victor@7 | 121 | data=upload_data, |
victor@7 | 122 | files=files, |
victor@7 | 123 | headers={'X-CSRFToken': upload_page.cookies['csrftoken'], |
moritz@29 | 124 | 'referer': self.upload_url}) |
victor@7 | 125 | |
victor@7 | 126 | if upload_submit.status_code != 200: |
i@14 | 127 | logger.warning("Connection error. Status code: %s" % upload_submit.status_code) |
victor@7 | 128 | return False |
victor@7 | 129 | |
victor@7 | 130 | # Check if there was a redirect to a new page. |
i@14 | 131 | if upload_submit.url == self.upload_url: |
victor@7 | 132 | measurement_id = False |
ioannis@32 | 133 | logger.error("Uploaded file(s) rejected! Try to upload manually to see the error.") |
victor@7 | 134 | else: |
victor@7 | 135 | measurement_id = re.findall(regex, upload_submit.text)[0] |
madrouin@23 | 136 | logger.info("Successfully uploaded measurement with id %s." % measurement_id) |
victor@7 | 137 | |
victor@7 | 138 | return measurement_id |
victor@7 | 139 | |
victor@7 | 140 | def download_files(self, measurement_id, subdir, download_url): |
victor@7 | 141 | """ Downloads some files from the download_url to the specified |
madrouin@20 | 142 | subdir. This method is used to download preprocessed file, optical |
victor@7 | 143 | files etc. |
victor@7 | 144 | """ |
ioannis@32 | 145 | # TODO: Make downloading more robust (e.g. in case that files do not exist on server). |
victor@7 | 146 | # Get the file |
moritz@29 | 147 | request = self.session.get(download_url, stream=True) |
moritz@29 | 148 | |
moritz@29 | 149 | if not request.ok: |
moritz@29 | 150 | raise Exception("Could not download files for measurement '%s'" % measurement_id) |
victor@7 | 151 | |
victor@7 | 152 | # Create the dir if it does not exist |
victor@7 | 153 | local_dir = os.path.join(self.output_dir, measurement_id, subdir) |
victor@7 | 154 | if not os.path.exists(local_dir): |
victor@7 | 155 | os.makedirs(local_dir) |
victor@7 | 156 | |
victor@7 | 157 | # Save the file by chunk, needed if the file is big. |
ioannis@32 | 158 | memory_file = BytesIO() |
victor@7 | 159 | |
victor@7 | 160 | for chunk in request.iter_content(chunk_size=1024): |
victor@7 | 161 | if chunk: # filter out keep-alive new chunks |
victor@7 | 162 | memory_file.write(chunk) |
victor@7 | 163 | memory_file.flush() |
victor@7 | 164 | |
victor@7 | 165 | zip_file = ZipFile(memory_file) |
victor@7 | 166 | |
victor@7 | 167 | for ziped_name in zip_file.namelist(): |
victor@7 | 168 | basename = os.path.basename(ziped_name) |
victor@7 | 169 | |
victor@7 | 170 | local_file = os.path.join(local_dir, basename) |
victor@7 | 171 | |
victor@7 | 172 | with open(local_file, 'wb') as f: |
victor@7 | 173 | f.write(zip_file.read(ziped_name)) |
victor@7 | 174 | |
ioannis@34 | 175 | def download_hirelpp(self, measurement_id): |
ioannis@34 | 176 | """ Download HiRElPP files for the measurement id. """ |
ioannis@34 | 177 | # Construct the download url |
ioannis@34 | 178 | download_url = self.download_hirelpp_pattern.format(measurement_id) |
ioannis@34 | 179 | self.download_files(measurement_id, 'hirelpp', download_url) |
ioannis@34 | 180 | |
ioannis@34 | 181 | def download_cloudmask(self, measurement_id): |
ioannis@34 | 182 | """ Download preprocessed files for the measurement id. """ |
ioannis@34 | 183 | # Construct the download url |
ioannis@34 | 184 | download_url = self.download_cloudmask_pattern.format(measurement_id) |
ioannis@34 | 185 | self.download_files(measurement_id, 'cloudmask', download_url) |
ioannis@34 | 186 | |
victor@7 | 187 | def download_preprocessed(self, measurement_id): |
victor@7 | 188 | """ Download preprocessed files for the measurement id. """ |
victor@7 | 189 | # Construct the download url |
i@14 | 190 | download_url = self.download_preprocessed_pattern.format(measurement_id) |
victor@7 | 191 | self.download_files(measurement_id, 'scc_preprocessed', download_url) |
victor@7 | 192 | |
victor@7 | 193 | def download_optical(self, measurement_id): |
victor@7 | 194 | """ Download optical files for the measurement id. """ |
victor@7 | 195 | # Construct the download url |
moritz@29 | 196 | download_url = self.download_optical_pattern.format(measurement_id) |
victor@7 | 197 | self.download_files(measurement_id, 'scc_optical', download_url) |
victor@7 | 198 | |
victor@7 | 199 | def download_graphs(self, measurement_id): |
victor@7 | 200 | """ Download profile graphs for the measurement id. """ |
victor@7 | 201 | # Construct the download url |
i@14 | 202 | download_url = self.download_graph_pattern.format(measurement_id) |
victor@7 | 203 | self.download_files(measurement_id, 'scc_plots', download_url) |
victor@7 | 204 | |
ioannis@34 | 205 | def download_elic(self, measurement_id): |
ioannis@34 | 206 | """ Download profile graphs for the measurement id. """ |
ioannis@34 | 207 | # Construct the download url |
ioannis@34 | 208 | download_url = self.download_elic_pattern.format(measurement_id) |
ioannis@34 | 209 | self.download_files(measurement_id, 'elic', download_url) |
ioannis@32 | 210 | |
victor@7 | 211 | def rerun_processing(self, measurement_id, monitor=True): |
madrouin@26 | 212 | measurement, status = self.get_measurement(measurement_id) |
victor@7 | 213 | |
victor@7 | 214 | if measurement: |
moritz@29 | 215 | request = self.session.get(measurement.rerun_processing_url, stream=True) |
victor@7 | 216 | |
victor@7 | 217 | if request.status_code != 200: |
i@14 | 218 | logger.error( |
i@14 | 219 | "Could not rerun processing for %s. Status code: %s" % (measurement_id, request.status_code)) |
victor@7 | 220 | return |
victor@7 | 221 | |
victor@7 | 222 | if monitor: |
victor@7 | 223 | self.monitor_processing(measurement_id) |
victor@7 | 224 | |
victor@7 | 225 | def rerun_all(self, measurement_id, monitor=True): |
victor@7 | 226 | logger.debug("Started rerun_all procedure.") |
victor@7 | 227 | |
victor@7 | 228 | logger.debug("Getting measurement %s" % measurement_id) |
madrouin@26 | 229 | measurement, status = self.get_measurement(measurement_id) |
victor@7 | 230 | |
victor@7 | 231 | if measurement: |
victor@7 | 232 | logger.debug("Attempting to rerun all processing through %s." % measurement.rerun_all_url) |
victor@7 | 233 | |
moritz@29 | 234 | request = self.session.get(measurement.rerun_all_url, stream=True) |
victor@7 | 235 | |
victor@7 | 236 | if request.status_code != 200: |
victor@7 | 237 | logger.error("Could not rerun pre processing for %s. Status code: %s" % |
victor@7 | 238 | (measurement_id, request.status_code)) |
victor@7 | 239 | return |
victor@7 | 240 | |
victor@7 | 241 | if monitor: |
victor@7 | 242 | self.monitor_processing(measurement_id) |
victor@7 | 243 | |
ioannis@32 | 244 | def process(self, filename, system_id, monitor, rs_filename=None, lr_filename=None, ov_filename=None): |
victor@7 | 245 | """ Upload a file for processing and wait for the processing to finish. |
victor@7 | 246 | If the processing is successful, it will download all produced files. |
victor@7 | 247 | """ |
victor@7 | 248 | logger.info("--- Processing started on %s. ---" % datetime.datetime.now()) |
victor@7 | 249 | # Upload file |
madrouin@26 | 250 | logger.info("--- Uploading file") |
ioannis@32 | 251 | measurement_id = self.upload_file(filename, system_id, |
ioannis@32 | 252 | rs_filename=rs_filename, |
ioannis@32 | 253 | lr_filename=lr_filename, |
ioannis@32 | 254 | ov_filename=ov_filename) |
victor@7 | 255 | |
ioannis@32 | 256 | if measurement_id and monitor: |
ioannis@32 | 257 | logger.info("--- Monitoring processing") |
moritz@29 | 258 | return self.monitor_processing(measurement_id) |
ioannis@32 | 259 | |
moritz@29 | 260 | return None |
victor@7 | 261 | |
victor@7 | 262 | def monitor_processing(self, measurement_id): |
victor@7 | 263 | """ Monitor the processing progress of a measurement id""" |
victor@7 | 264 | |
madrouin@24 | 265 | # try to deal with error 404 |
madrouin@24 | 266 | error_count = 0 |
madrouin@24 | 267 | error_max = 6 |
madrouin@26 | 268 | time_sleep = 10 |
madrouin@24 | 269 | |
madrouin@25 | 270 | # try to wait for measurement to appear in API |
madrouin@24 | 271 | measurement = None |
ioannis@32 | 272 | logger.info("Looking for measurement %s in SCC", measurement_id) |
madrouin@26 | 273 | while error_count < error_max: |
madrouin@26 | 274 | time.sleep(time_sleep) |
madrouin@26 | 275 | measurement, status = self.get_measurement(measurement_id) |
madrouin@26 | 276 | if status != 200 and error_count < error_max: |
ioannis@32 | 277 | logger.error("Measurement not found. waiting %ds", time_sleep) |
madrouin@24 | 278 | error_count += 1 |
madrouin@26 | 279 | else: |
madrouin@26 | 280 | break |
madrouin@24 | 281 | |
madrouin@24 | 282 | if error_count == error_max: |
ioannis@32 | 283 | logger.critical("Measurement %s doesn't seem to exist", measurement_id) |
madrouin@26 | 284 | sys.exit(1) |
madrouin@26 | 285 | |
ioannis@32 | 286 | logger.info('Measurement %s found', measurement_id) |
madrouin@24 | 287 | |
victor@7 | 288 | if measurement is not None: |
victor@7 | 289 | while measurement.is_running: |
ioannis@34 | 290 | logger.info("Measurement is being processed. Please wait.") |
victor@7 | 291 | time.sleep(10) |
madrouin@26 | 292 | measurement, status = self.get_measurement(measurement_id) |
madrouin@24 | 293 | |
ioannis@34 | 294 | logger.info("Measurement processing finished.") |
ioannis@34 | 295 | if measurement.hirelpp == 127: |
ioannis@34 | 296 | logger.info("Downloading hirelpp files.") |
ioannis@34 | 297 | self.download_hirelpp(measurement_id) |
ioannis@34 | 298 | if measurement.cloudmask == 127: |
ioannis@34 | 299 | logger.info("Downloading cloudmask files.") |
ioannis@34 | 300 | self.download_cloudmask(measurement_id) |
ioannis@34 | 301 | if measurement.elpp == 127: |
victor@7 | 302 | logger.info("Downloading preprocessed files.") |
victor@7 | 303 | self.download_preprocessed(measurement_id) |
ioannis@34 | 304 | if measurement.elda == 127: |
victor@7 | 305 | logger.info("Downloading optical files.") |
victor@7 | 306 | self.download_optical(measurement_id) |
victor@7 | 307 | logger.info("Downloading graphs.") |
victor@7 | 308 | self.download_graphs(measurement_id) |
ioannis@34 | 309 | if measurement.elic == 127: |
ioannis@34 | 310 | logger.info("Downloading preprocessed files.") |
ioannis@34 | 311 | self.download_elic(measurement_id) |
victor@7 | 312 | logger.info("--- Processing finished. ---") |
victor@7 | 313 | return measurement |
victor@7 | 314 | |
victor@7 | 315 | def get_measurement(self, measurement_id): |
i@31 | 316 | measurement_url = self.api_measurement_pattern.format(measurement_id) |
i@31 | 317 | logger.debug("Measurement API URL: %s" % measurement_url) |
victor@7 | 318 | |
moritz@29 | 319 | response = self.session.get(measurement_url) |
victor@7 | 320 | |
moritz@29 | 321 | if not response.ok: |
i@14 | 322 | logger.error('Could not access API. Status code %s.' % response.status_code) |
madrouin@26 | 323 | return None, response.status_code |
i@14 | 324 | |
victor@7 | 325 | response_dict = response.json() |
victor@7 | 326 | |
victor@7 | 327 | if response_dict: |
moritz@29 | 328 | measurement = Measurement(self.base_url, response_dict) |
madrouin@26 | 329 | return measurement, response.status_code |
victor@7 | 330 | else: |
victor@7 | 331 | logger.error("No measurement with id %s found on the SCC." % measurement_id) |
madrouin@26 | 332 | return None, response.status_code |
victor@7 | 333 | |
victor@7 | 334 | def delete_measurement(self, measurement_id): |
victor@7 | 335 | """ Deletes a measurement with the provided measurement id. The user |
madrouin@20 | 336 | should have the appropriate permissions. |
madrouin@20 | 337 | |
victor@7 | 338 | The procedures is performed directly through the web interface and |
victor@7 | 339 | NOT through the API. |
victor@7 | 340 | """ |
victor@7 | 341 | # Get the measurement object |
i@31 | 342 | measurement, _ = self.get_measurement(measurement_id) |
victor@7 | 343 | |
victor@7 | 344 | # Check that it exists |
victor@7 | 345 | if measurement is None: |
victor@7 | 346 | logger.warning("Nothing to delete.") |
victor@7 | 347 | return None |
victor@7 | 348 | |
victor@7 | 349 | # Go the the page confirming the deletion |
moritz@29 | 350 | delete_url = self.delete_measurement_pattern.format(measurement_id) |
victor@7 | 351 | |
moritz@29 | 352 | confirm_page = self.session.get(delete_url) |
victor@7 | 353 | |
victor@7 | 354 | # Check that the page opened properly |
victor@7 | 355 | if confirm_page.status_code != 200: |
victor@7 | 356 | logger.warning("Could not open delete page. Status: {0}".format(confirm_page.status_code)) |
victor@7 | 357 | return None |
victor@7 | 358 | |
victor@7 | 359 | # Delete the measurement |
victor@7 | 360 | delete_page = self.session.post(delete_url, |
victor@7 | 361 | data={'post': 'yes'}, |
victor@7 | 362 | headers={'X-CSRFToken': confirm_page.cookies['csrftoken'], |
victor@7 | 363 | 'referer': delete_url} |
victor@7 | 364 | ) |
i@31 | 365 | if not delete_page.ok: |
victor@7 | 366 | logger.warning("Something went wrong. Delete page status: {0}".format( |
victor@7 | 367 | delete_page.status_code)) |
victor@7 | 368 | return None |
victor@7 | 369 | |
victor@7 | 370 | logger.info("Deleted measurement {0}".format(measurement_id)) |
victor@7 | 371 | return True |
victor@7 | 372 | |
victor@7 | 373 | def available_measurements(self): |
victor@7 | 374 | """ Get a list of available measurement on the SCC. """ |
i@31 | 375 | response = self.session.get(self.api_measurements_url) |
victor@7 | 376 | response_dict = response.json() |
victor@7 | 377 | |
victor@7 | 378 | if response_dict: |
victor@7 | 379 | measurement_list = response_dict['objects'] |
i@14 | 380 | measurements = [Measurement(self.base_url, measurement_dict) for measurement_dict in measurement_list] |
victor@7 | 381 | logger.info("Found %s measurements on the SCC." % len(measurements)) |
victor@7 | 382 | else: |
victor@7 | 383 | logger.warning("No response received from the SCC when asked for available measurements.") |
i@31 | 384 | measurements = None |
victor@7 | 385 | |
victor@7 | 386 | return measurements |
victor@7 | 387 | |
moritz@29 | 388 | def list_measurements(self, station=None, system=None, start=None, stop=None, upload_status=None, |
moritz@29 | 389 | processing_status=None, optical_processing=None): |
moritz@29 | 390 | |
i@31 | 391 | # TODO: Change this to work through the API |
i@31 | 392 | |
moritz@29 | 393 | # Need to set to empty string if not specified, we won't get any results |
moritz@29 | 394 | params = { |
moritz@29 | 395 | "station": station if station is not None else "", |
moritz@29 | 396 | "system": system if system is not None else "", |
moritz@29 | 397 | "stop": stop if stop is not None else "", |
moritz@29 | 398 | "start": start if start is not None else "", |
moritz@29 | 399 | "upload_status": upload_status if upload_status is not None else "", |
moritz@29 | 400 | "preprocessing_status": processing_status if processing_status is not None else "", |
moritz@29 | 401 | "optical_processing_status": optical_processing if optical_processing is not None else "" |
moritz@29 | 402 | } |
i@31 | 403 | |
i@31 | 404 | response_txt = self.session.get(self.list_measurements_url, params=params).text |
moritz@29 | 405 | tbl_rgx = re.compile(r'<table id="measurements">(.*?)</table>', re.DOTALL) |
moritz@29 | 406 | entry_rgx = re.compile(r'<tr>(.*?)</tr>', re.DOTALL) |
moritz@29 | 407 | measurement_rgx = re.compile( |
moritz@29 | 408 | r'.*?<td><a[^>]*>(\w+)</a>.*?<td>.*?<td>([\w-]+ [\w:]+)</td>.*<td data-order="([-]?\d+),([-]?\d+),([-]?\d+)".*', |
moritz@29 | 409 | re.DOTALL) |
i@31 | 410 | matches = tbl_rgx.findall(response_txt) |
moritz@29 | 411 | if len(matches) != 1: |
moritz@29 | 412 | return [] |
moritz@29 | 413 | |
moritz@29 | 414 | ret = [] |
moritz@29 | 415 | for entry in entry_rgx.finditer(matches[0]): |
moritz@29 | 416 | m = measurement_rgx.match(entry.string[entry.start(0):entry.end(0)]) |
moritz@29 | 417 | if m: |
moritz@29 | 418 | name, date, upload, preproc, optical = m.groups() |
moritz@29 | 419 | ret.append( |
moritz@29 | 420 | Measurement(self.base_url, {"id": name, "upload": int(upload), "pre_processing": int(preproc), |
moritz@29 | 421 | "processing": int(optical)})) |
moritz@29 | 422 | |
moritz@29 | 423 | return ret |
moritz@29 | 424 | |
i@31 | 425 | def measurement_id_for_date(self, t1, call_sign, base_number=0): |
victor@7 | 426 | """ Give the first available measurement id on the SCC for the specific |
madrouin@20 | 427 | date. |
victor@7 | 428 | """ |
victor@7 | 429 | date_str = t1.strftime('%Y%m%d') |
i@31 | 430 | base_id = "%s%s" % (date_str, call_sign) |
i@31 | 431 | search_url = urlparse.urljoin(self.api_base_url, 'measurements/?id__startswith=%s' % base_id) |
victor@7 | 432 | |
moritz@29 | 433 | response = self.session.get(search_url) |
victor@7 | 434 | |
victor@7 | 435 | response_dict = response.json() |
victor@7 | 436 | |
victor@7 | 437 | measurement_id = None |
victor@7 | 438 | |
victor@7 | 439 | if response_dict: |
victor@7 | 440 | measurement_list = response_dict['objects'] |
i@31 | 441 | |
i@31 | 442 | if len(measurement_list) == 100: |
i@31 | 443 | raise ValueError('No available measurement id found.') |
i@31 | 444 | |
victor@7 | 445 | existing_ids = [measurement_dict['id'] for measurement_dict in measurement_list] |
victor@7 | 446 | |
victor@7 | 447 | measurement_number = base_number |
i@31 | 448 | measurement_id = "%s%02i" % (base_id, measurement_number) |
victor@7 | 449 | |
victor@7 | 450 | while measurement_id in existing_ids: |
victor@7 | 451 | measurement_number = measurement_number + 1 |
i@31 | 452 | measurement_id = "%s%02i" % (base_id, measurement_number) |
victor@7 | 453 | |
victor@7 | 454 | return measurement_id |
victor@7 | 455 | |
i@31 | 456 | class PageNotAccessibleError(RuntimeError): |
i@31 | 457 | pass |
i@31 | 458 | |
victor@7 | 459 | |
ioannis@17 | 460 | class ApiObject(object): |
victor@7 | 461 | """ A generic class object. """ |
victor@7 | 462 | |
i@14 | 463 | def __init__(self, base_url, dict_response): |
i@14 | 464 | self.base_url = base_url |
victor@7 | 465 | |
victor@7 | 466 | if dict_response: |
victor@7 | 467 | # Add the dictionary key value pairs as object properties |
victor@7 | 468 | for key, value in dict_response.items(): |
ioannis@17 | 469 | # logger.debug('Setting key {0} to value {1}'.format(key, value)) |
ioannis@17 | 470 | try: |
ioannis@17 | 471 | setattr(self, key, value) |
ioannis@17 | 472 | except: |
ioannis@17 | 473 | logger.warning('Could not set attribute {0} to value {1}'.format(key, value)) |
victor@7 | 474 | self.exists = True |
victor@7 | 475 | else: |
victor@7 | 476 | self.exists = False |
victor@7 | 477 | |
victor@7 | 478 | |
victor@7 | 479 | class Measurement(ApiObject): |
victor@7 | 480 | """ This class represents the measurement object as returned in the SCC API. |
victor@7 | 481 | """ |
victor@7 | 482 | |
victor@7 | 483 | @property |
victor@7 | 484 | def rerun_processing_url(self): |
ioannis@34 | 485 | url_pattern = urlparse.urljoin(self.base_url, 'data_processing/measurements/{0}/rerun-elda/') |
i@14 | 486 | return url_pattern.format(self.id) |
victor@7 | 487 | |
victor@7 | 488 | @property |
victor@7 | 489 | def rerun_all_url(self): |
i@14 | 490 | ulr_pattern = urlparse.urljoin(self.base_url, 'data_processing/measurements/{0}/rerun-all/') |
i@14 | 491 | return ulr_pattern.format(self.id) |
victor@7 | 492 | |
victor@7 | 493 | def __str__(self): |
victor@7 | 494 | return "%s: %s, %s, %s" % (self.id, |
victor@7 | 495 | self.upload, |
victor@7 | 496 | self.pre_processing, |
victor@7 | 497 | self.processing) |
victor@7 | 498 | |
victor@7 | 499 | |
ioannis@32 | 500 | def process_file(filename, system_id, settings, monitor=True, rs_filename=None, lr_filename=None, ov_filename=None): |
ioannis@32 | 501 | """ Shortcut function to process a file to the SCC. """ |
ioannis@32 | 502 | logger.info("Processing file %s, using system %s" % (filename, system_id)) |
victor@7 | 503 | |
i@14 | 504 | scc = SCC(settings['basic_credentials'], settings['output_dir'], settings['base_url']) |
i@14 | 505 | scc.login(settings['website_credentials']) |
ioannis@32 | 506 | measurement = scc.process(filename, system_id, |
ioannis@32 | 507 | monitor=monitor, |
ioannis@32 | 508 | rs_filename=rs_filename, |
ioannis@32 | 509 | lr_filename=lr_filename, |
ioannis@32 | 510 | ov_filename=ov_filename) |
victor@7 | 511 | scc.logout() |
victor@7 | 512 | return measurement |
victor@7 | 513 | |
victor@7 | 514 | |
ioannis@32 | 515 | def delete_measurements(measurement_ids, settings): |
moritz@29 | 516 | """ Shortcut function to delete measurements from the SCC. """ |
moritz@29 | 517 | scc = SCC(settings['basic_credentials'], settings['output_dir'], settings['base_url']) |
moritz@29 | 518 | scc.login(settings['website_credentials']) |
moritz@29 | 519 | for m_id in measurement_ids: |
moritz@29 | 520 | logger.info("Deleting %s" % m_id) |
moritz@29 | 521 | scc.delete_measurement(m_id) |
moritz@29 | 522 | scc.logout() |
moritz@29 | 523 | |
moritz@29 | 524 | |
moritz@29 | 525 | def rerun_all(measurement_ids, monitor, settings): |
moritz@29 | 526 | """ Shortcut function to rerun measurements from the SCC. """ |
i@14 | 527 | |
i@14 | 528 | scc = SCC(settings['basic_credentials'], settings['output_dir'], settings['base_url']) |
i@14 | 529 | scc.login(settings['website_credentials']) |
moritz@29 | 530 | for m_id in measurement_ids: |
moritz@29 | 531 | logger.info("Rerunning all products for %s" % m_id) |
moritz@29 | 532 | scc.rerun_all(m_id, monitor) |
victor@7 | 533 | scc.logout() |
victor@7 | 534 | |
victor@7 | 535 | |
moritz@29 | 536 | def rerun_processing(measurement_ids, monitor, settings): |
victor@7 | 537 | """ Shortcut function to delete a measurement from the SCC. """ |
i@14 | 538 | |
i@14 | 539 | scc = SCC(settings['basic_credentials'], settings['output_dir'], settings['base_url']) |
i@14 | 540 | scc.login(settings['website_credentials']) |
moritz@29 | 541 | for m_id in measurement_ids: |
moritz@29 | 542 | logger.info("Rerunning (optical) processing for %s" % m_id) |
moritz@29 | 543 | scc.rerun_processing(m_id, monitor) |
moritz@29 | 544 | scc.logout() |
moritz@29 | 545 | |
moritz@29 | 546 | |
moritz@29 | 547 | def list_measurements(settings, station=None, system=None, start=None, stop=None, upload_status=None, |
moritz@29 | 548 | preprocessing_status=None, |
moritz@29 | 549 | optical_processing=None): |
moritz@29 | 550 | """List all available measurements""" |
moritz@29 | 551 | scc = SCC(settings['basic_credentials'], settings['output_dir'], settings['base_url']) |
moritz@29 | 552 | scc.login(settings['website_credentials']) |
moritz@29 | 553 | ret = scc.list_measurements(station=station, system=system, start=start, stop=stop, upload_status=upload_status, |
moritz@29 | 554 | processing_status=preprocessing_status, optical_processing=optical_processing) |
moritz@29 | 555 | for entry in ret: |
moritz@29 | 556 | print("%s" % entry.id) |
victor@7 | 557 | scc.logout() |
victor@7 | 558 | |
victor@7 | 559 | |
moritz@29 | 560 | def download_measurements(measurement_ids, download_preproc, download_optical, download_graph, settings): |
moritz@29 | 561 | """Download all measurements for the specified IDs""" |
i@14 | 562 | scc = SCC(settings['basic_credentials'], settings['output_dir'], settings['base_url']) |
i@14 | 563 | scc.login(settings['website_credentials']) |
moritz@29 | 564 | for m_id in measurement_ids: |
moritz@29 | 565 | if download_preproc: |
moritz@29 | 566 | logger.info("Downloading preprocessed files for '%s'" % m_id) |
moritz@29 | 567 | scc.download_preprocessed(m_id) |
moritz@29 | 568 | logger.info("Complete") |
moritz@29 | 569 | if download_optical: |
moritz@29 | 570 | logger.info("Downloading optical files for '%s'" % m_id) |
moritz@29 | 571 | scc.download_optical(m_id) |
moritz@29 | 572 | logger.info("Complete") |
moritz@29 | 573 | if download_graph: |
moritz@29 | 574 | logger.info("Downloading profile graph files for '%s'" % m_id) |
moritz@29 | 575 | scc.download_graphs(m_id) |
moritz@29 | 576 | logger.info("Complete") |
i@14 | 577 | |
i@14 | 578 | |
moritz@29 | 579 | def settings_from_path(config_file_path): |
i@14 | 580 | """ Read the configuration file. |
i@14 | 581 | |
i@14 | 582 | The file should be in YAML syntax.""" |
i@14 | 583 | |
i@14 | 584 | if not os.path.isfile(config_file_path): |
moritz@29 | 585 | raise argparse.ArgumentTypeError("Wrong path for configuration file (%s)" % config_file_path) |
i@14 | 586 | |
i@14 | 587 | with open(config_file_path) as yaml_file: |
i@14 | 588 | try: |
i@14 | 589 | settings = yaml.safe_load(yaml_file) |
i@14 | 590 | logger.debug("Read settings file(%s)" % config_file_path) |
moritz@29 | 591 | except Exception: |
moritz@29 | 592 | raise argparse.ArgumentTypeError("Could not parse YAML file (%s)" % config_file_path) |
i@14 | 593 | |
i@14 | 594 | # YAML limitation: does not read tuples |
i@14 | 595 | settings['basic_credentials'] = tuple(settings['basic_credentials']) |
i@14 | 596 | settings['website_credentials'] = tuple(settings['website_credentials']) |
i@14 | 597 | return settings |
i@14 | 598 | |
i@14 | 599 | |
moritz@29 | 600 | # Setup for command specific parsers |
moritz@29 | 601 | def setup_delete(parser): |
moritz@29 | 602 | def delete_from_args(parsed): |
ioannis@32 | 603 | delete_measurements(parsed.IDs, parsed.config) |
moritz@29 | 604 | |
moritz@29 | 605 | parser.add_argument("IDs", nargs="+", help="measurement IDs to delete.") |
moritz@29 | 606 | parser.set_defaults(execute=delete_from_args) |
moritz@29 | 607 | |
moritz@29 | 608 | |
moritz@29 | 609 | def setup_rerun_all(parser): |
moritz@29 | 610 | def rerun_all_from_args(parsed): |
moritz@29 | 611 | rerun_all(parsed.IDs, parsed.process, parsed.config) |
moritz@29 | 612 | |
moritz@29 | 613 | parser.add_argument("IDs", nargs="+", help="Measurement IDs to rerun.") |
moritz@29 | 614 | parser.add_argument("-p", "--process", help="Wait for the results of the processing.", |
moritz@29 | 615 | action="store_true") |
moritz@29 | 616 | parser.set_defaults(execute=rerun_all_from_args) |
moritz@29 | 617 | |
moritz@29 | 618 | |
moritz@29 | 619 | def setup_rerun_processing(parser): |
moritz@29 | 620 | def rerun_processing_from_args(parsed): |
moritz@29 | 621 | rerun_processing(parsed.IDs, parsed.process, parsed.config) |
moritz@29 | 622 | |
moritz@29 | 623 | parser.add_argument("IDs", nargs="+", help="Measurement IDs to rerun the processing on.") |
moritz@29 | 624 | parser.add_argument("-p", "--process", help="Wait for the results of the processing.", |
moritz@29 | 625 | action="store_true") |
moritz@29 | 626 | parser.set_defaults(execute=rerun_processing_from_args) |
moritz@29 | 627 | |
moritz@29 | 628 | |
moritz@29 | 629 | def setup_process_file(parser): |
ioannis@32 | 630 | """ Upload and monitor processing progress.""" |
moritz@29 | 631 | def process_file_from_args(parsed): |
ioannis@32 | 632 | process_file(parsed.filename, parsed.system, parsed.config, monitor=True, |
ioannis@32 | 633 | rs_filename=parsed.radiosounding, |
ioannis@32 | 634 | ov_filename=parsed.overlap, |
ioannis@32 | 635 | lr_filename=parsed.lidarratio) |
moritz@29 | 636 | |
moritz@29 | 637 | parser.add_argument("filename", help="Measurement file name or path.") |
moritz@29 | 638 | parser.add_argument("system", help="Processing system id.") |
i@30 | 639 | parser.add_argument("--radiosounding", default=None, help="Radiosounding file name or path") |
ioannis@32 | 640 | parser.add_argument("--overlap", default=None, help="Overlap file name or path") |
ioannis@32 | 641 | parser.add_argument("--lidarratio", default=None, help="Lidar ratio file name or path") |
ioannis@32 | 642 | |
moritz@29 | 643 | parser.set_defaults(execute=process_file_from_args) |
moritz@29 | 644 | |
moritz@29 | 645 | |
moritz@29 | 646 | def setup_upload_file(parser): |
ioannis@32 | 647 | """ Upload but do not monitor processing progress. """ |
moritz@29 | 648 | def upload_file_from_args(parsed): |
ioannis@32 | 649 | process_file(parsed.filename, parsed.system, parsed.config, monitor=False, |
ioannis@32 | 650 | rs_filename=parsed.radiosounding, |
ioannis@32 | 651 | ov_filename=parsed.overlap, |
ioannis@32 | 652 | lr_filename=parsed.lidarratio) |
moritz@29 | 653 | |
moritz@29 | 654 | parser.add_argument("filename", help="Measurement file name or path.") |
moritz@29 | 655 | parser.add_argument("system", help="Processing system id.") |
i@30 | 656 | parser.add_argument("--radiosounding", default=None, help="Radiosounding file name or path") |
ioannis@32 | 657 | parser.add_argument("--overlap", default=None, help="Overlap file name or path") |
ioannis@32 | 658 | parser.add_argument("--lidarratio", default=None, help="Lidar ratio file name or path") |
i@30 | 659 | |
moritz@29 | 660 | parser.set_defaults(execute=upload_file_from_args) |
moritz@29 | 661 | |
moritz@29 | 662 | |
moritz@29 | 663 | def setup_list_measurements(parser): |
moritz@29 | 664 | def list_measurements_from_args(parsed): |
moritz@29 | 665 | list_measurements(parsed.config, station=parsed.station, system=parsed.system, start=parsed.start, |
moritz@29 | 666 | stop=parsed.stop, |
moritz@29 | 667 | upload_status=parsed.upload_status, preprocessing_status=parsed.preprocessing_status, |
moritz@29 | 668 | optical_processing=parsed.optical_processing_status) |
moritz@29 | 669 | |
moritz@29 | 670 | def status(arg): |
moritz@29 | 671 | if -127 <= int(arg) <= 127: |
moritz@29 | 672 | return arg |
moritz@29 | 673 | else: |
moritz@29 | 674 | raise argparse.ArgumentTypeError("Status must be between -127 and 127") |
moritz@29 | 675 | |
moritz@29 | 676 | def date(arg): |
moritz@29 | 677 | if re.match(r'\d{4}-\d{2}-\d{2}', arg): |
moritz@29 | 678 | return arg |
moritz@29 | 679 | else: |
moritz@29 | 680 | raise argparse.ArgumentTypeError("Date must be in format 'YYYY-MM-DD'") |
moritz@29 | 681 | |
moritz@29 | 682 | parser.add_argument("--station", help="Filter for only the selected station") |
moritz@29 | 683 | parser.add_argument("--system", help="Filter for only the selected station") |
moritz@29 | 684 | parser.add_argument("--start", help="Filter for only the selected station", type=date) |
moritz@29 | 685 | parser.add_argument("--stop", help="Filter for only the selected station", type=date) |
moritz@29 | 686 | parser.add_argument("--upload-status", help="Filter for only the selected station", type=status) |
moritz@29 | 687 | parser.add_argument("--preprocessing-status", help="Filter for only the selected station", type=status) |
moritz@29 | 688 | parser.add_argument("--optical-processing-status", help="Filter for only the selected station", type=status) |
moritz@29 | 689 | parser.set_defaults(execute=list_measurements_from_args) |
moritz@29 | 690 | |
moritz@29 | 691 | |
moritz@29 | 692 | def setup_download_measurements(parser): |
moritz@29 | 693 | def download_measurements_from_args(parsed): |
moritz@29 | 694 | preproc = parsed.download_preprocessed |
moritz@29 | 695 | optical = parsed.download_optical |
moritz@29 | 696 | graphs = parsed.download_profile_graphs |
moritz@29 | 697 | if not preproc and not graphs: |
moritz@29 | 698 | optical = True |
moritz@29 | 699 | download_measurements(parsed.IDs, preproc, optical, graphs, parsed.config) |
moritz@29 | 700 | |
moritz@29 | 701 | parser.add_argument("IDs", help="Measurement IDs that should be downloaded.", nargs="+") |
moritz@29 | 702 | parser.add_argument("--download-preprocessed", action="store_true", help="Download preprocessed files.") |
moritz@29 | 703 | parser.add_argument("--download-optical", action="store_true", |
moritz@29 | 704 | help="Download optical files (default if no other download is used).") |
moritz@29 | 705 | parser.add_argument("--download-profile-graphs", action="store_true", help="Download profile graph files.") |
moritz@29 | 706 | parser.set_defaults(execute=download_measurements_from_args) |
moritz@29 | 707 | |
moritz@29 | 708 | |
victor@7 | 709 | def main(): |
victor@7 | 710 | # Define the command line arguments. |
victor@7 | 711 | parser = argparse.ArgumentParser() |
moritz@29 | 712 | subparsers = parser.add_subparsers() |
moritz@29 | 713 | |
moritz@29 | 714 | delete_parser = subparsers.add_parser("delete", help="Deletes a measurement.") |
ioannis@32 | 715 | rerun_all_parser = subparsers.add_parser("rerun-all", help="Reprocess a measurement on the SCC.") |
moritz@29 | 716 | rerun_processing_parser = subparsers.add_parser("rerun-processing", |
moritz@29 | 717 | help="Rerun processing routings for a measurement.") |
ioannis@32 | 718 | process_file_parser = subparsers.add_parser("process-file", help="Upload a file and download procesing results.") |
moritz@29 | 719 | upload_file_parser = subparsers.add_parser("upload-file", help="Upload a file.") |
ioannis@32 | 720 | list_parser = subparsers.add_parser("list", help="List measurements registered on the SCC.") |
moritz@29 | 721 | download_parser = subparsers.add_parser("download", help="Download selected measurements.") |
moritz@29 | 722 | |
moritz@29 | 723 | setup_delete(delete_parser) |
moritz@29 | 724 | setup_rerun_all(rerun_all_parser) |
moritz@29 | 725 | setup_rerun_processing(rerun_processing_parser) |
moritz@29 | 726 | setup_process_file(process_file_parser) |
moritz@29 | 727 | setup_upload_file(upload_file_parser) |
moritz@29 | 728 | setup_list_measurements(list_parser) |
moritz@29 | 729 | setup_download_measurements(download_parser) |
victor@7 | 730 | |
victor@7 | 731 | # Verbosity settings from http://stackoverflow.com/a/20663028 |
victor@7 | 732 | parser.add_argument('-d', '--debug', help="Print debugging information.", action="store_const", |
victor@7 | 733 | dest="loglevel", const=logging.DEBUG, default=logging.INFO, |
victor@7 | 734 | ) |
victor@7 | 735 | parser.add_argument('-s', '--silent', help="Show only warning and error messages.", action="store_const", |
victor@7 | 736 | dest="loglevel", const=logging.WARNING |
victor@7 | 737 | ) |
victor@7 | 738 | |
ioannis@32 | 739 | # Setup default config location |
moritz@29 | 740 | home = os.path.expanduser("~") |
moritz@29 | 741 | default_config_location = os.path.abspath(os.path.join(home, ".scc_access.yaml")) |
moritz@29 | 742 | parser.add_argument("-c", "--config", help="Path to the config file.", type=settings_from_path, |
moritz@29 | 743 | default=default_config_location) |
moritz@29 | 744 | |
victor@7 | 745 | args = parser.parse_args() |
victor@7 | 746 | |
victor@7 | 747 | # Get the logger with the appropriate level |
victor@7 | 748 | logging.basicConfig(format='%(levelname)s: %(message)s', level=args.loglevel) |
victor@7 | 749 | |
moritz@29 | 750 | # Dispatch to appropriate function |
moritz@29 | 751 | args.execute(args) |
moritz@29 | 752 | |
i@14 | 753 | |
moritz@29 | 754 | # When running through terminal |
moritz@29 | 755 | if __name__ == '__main__': |
moritz@29 | 756 | main() |