infoex-autowx.py

   1 #!/usr/bin/env python
   2 # -*- coding: utf-8 -*-
   3
   4 """
   5 InfoEx <-> NRCS/MesoWest Auto Wx implementation
   6 Alexander Vasarab
   7 Wylark Mountaineering LLC
   8
   9 This program fetches data from either an NRCS SNOTEL site or MesoWest
  10 weather station and pushes it to InfoEx using the new automated weather
  11 system implementation.
  12
  13 It is designed to be run hourly, and it asks for the last three hours
  14 of data of each desired type, and selects the most recent one. This
  15 lends some resiliency to the process and helps ensure that we have a
  16 value to send, but it can lead to somewhat inconsistent/untruthful
  17 data if e.g. the HS is from the last hour but the tempPres is from two
  18 hours ago because the instrumentation had a hiccup. It's worth
  19 considering if this is a bug or a feature.
  20
  21 For more information, see file: README
  22 For licensing, see file: LICENSE
  23 """
  24
  25 import configparser
  26 import csv
  27 import datetime
  28 import logging
  29 import os
  30 import sys
  31 import time
  32
  33 from ftplib import FTP
  34 from argparse import ArgumentParser
  35
  36 import requests
  37
  38 import zeep
  39 import zeep.cache
  40 import zeep.transports
  41
  42 __version__ = '3.0.1'
  43
  44 LOG = logging.getLogger(__name__)
  45 LOG.setLevel(logging.NOTSET)
  46
  47 def get_parser():
  48     """Return OptionParser for this program"""
  49     parser = ArgumentParser()
  50
  51     parser.add_argument("--version",
  52                         action="version",
  53                         version=__version__)
  54
  55     parser.add_argument("--config",
  56                         dest="config",
  57                         metavar="FILE",
  58                         help="location of config file")
  59
  60     parser.add_argument("--log-level",
  61                         dest="log_level",
  62                         default=None,
  63                         help="set the log level (debug, info, warning)")
  64
  65     parser.add_argument("--dry-run",
  66                         action="store_true",
  67                         dest="dry_run",
  68                         default=False,
  69                         help="fetch data but don't upload to InfoEx")
  70
  71     return parser
  72
  73 def setup_config(config):
  74     """Setup config variable based on values specified in the ini file"""
  75     try:
  76         infoex = {
  77             'host': config['infoex']['host'],
  78             'uuid': config['infoex']['uuid'],
  79             'api_key': config['infoex']['api_key'],
  80             'csv_filename': config['infoex']['csv_filename'],
  81             'location_uuid': config['infoex']['location_uuid'],
  82             'wx_data': {}, # placeholder key, values to come later
  83         }
  84
  85         station = dict()
  86         station['provider'] = config['station']['type']
  87
  88         if station['provider'] not in ['nrcs', 'mesowest', 'python']:
  89             print("Please specify either nrcs or mesowest as the station type.")
  90             sys.exit(1)
  91
  92         if station['provider'] == 'nrcs':
  93             station['source'] = 'https://www.wcc.nrcs.usda.gov/awdbWebService/services?WSDL'
  94             station['station_id'] = config['station']['station_id']
  95             station['desired_data'] = config['station']['desired_data'].split(',')
  96
  97             # XXX: For NRCS, we're manually overriding units for now! Once
  98             #      unit conversion is supported for NRCS, REMOVE THIS!
  99             if 'units' not in station:
 100                 station['units'] = 'imperial'
 101
 102         if station['provider'] == 'mesowest':
 103             station['source'] = 'https://api.synopticdata.com/v2/stations/timeseries'
 104             station['station_id'] = config['station']['station_id']
 105             station['units'] = config['station']['units']
 106             station['desired_data'] = config['station']['desired_data']
 107
 108             # construct full API URL (sans start/end time, added later)
 109             station['source'] = station['source'] + '?token=' + \
 110                                 config['station']['token'] + \
 111                                 '&within=60&units=' + station['units'] + \
 112                                 '&stid=' + station['station_id'] + \
 113                                 '&vars=' + station['desired_data']
 114
 115         if station['provider'] == 'python':
 116             station['path'] = config['station']['path']
 117
 118     except KeyError as err:
 119         LOG.critical("%s not defined in configuration file", err)
 120         sys.exit(1)
 121
 122     # all sections/values present in config file, final sanity check
 123     try:
 124         for key in config.sections():
 125             for subkey in config[key]:
 126                 if not config[key][subkey]:
 127                     raise ValueError
 128     except ValueError:
 129         LOG.critical("Config value '%s.%s' is empty", key, subkey)
 130         sys.exit(1)
 131
 132     return (infoex, station)
 133
 134 def setup_logging(log_level):
 135     """Setup our logging infrastructure"""
 136     try:
 137         from systemd.journal import JournalHandler
 138         LOG.addHandler(JournalHandler())
 139     except ImportError:
 140         ## fallback to syslog
 141         #import logging.handlers
 142         #LOG.addHandler(logging.handlers.SysLogHandler())
 143         # fallback to stdout
 144         handler = logging.StreamHandler(sys.stdout)
 145         LOG.addHandler(handler)
 146
 147     # ugly, but passable
 148     if log_level in [None, 'debug', 'info', 'warning']:
 149         if log_level == 'debug':
 150             LOG.setLevel(logging.DEBUG)
 151         elif log_level == 'info':
 152             LOG.setLevel(logging.INFO)
 153         elif log_level == 'warning':
 154             LOG.setLevel(logging.WARNING)
 155         else:
 156             LOG.setLevel(logging.NOTSET)
 157     else:
 158         return False
 159
 160     return True
 161
 162 def main():
 163     """Main routine: sort through args, decide what to do, then do it"""
 164     parser = get_parser()
 165     options = parser.parse_args()
 166
 167     config = configparser.ConfigParser(allow_no_value=False)
 168
 169     if not options.config:
 170         parser.print_help()
 171         print("\nPlease specify a configuration file via --config.")
 172         sys.exit(1)
 173
 174     config.read(options.config)
 175
 176     if not setup_logging(options.log_level):
 177         parser.print_help()
 178         print("\nPlease select an appropriate log level or remove the switch (--log-level).")
 179         sys.exit(1)
 180
 181     (infoex, station) = setup_config(config)
 182
 183     LOG.debug('Config parsed, starting up')
 184
 185     # create mappings
 186     (fmap, final_data) = setup_infoex_fields_mapping(infoex['location_uuid'])
 187     iemap = setup_infoex_counterparts_mapping(station['provider'])
 188
 189     # override units if user selected metric
 190     try:
 191         if station['units'] == 'metric':
 192             final_data = switch_units_to_metric(final_data, fmap)
 193     except KeyError:
 194         if station['provider'] != 'python':
 195             LOG.error("Please specify the units in the configuration "
 196                       "file")
 197             sys.exit(1)
 198
 199     (begin_date, end_date) = setup_time_values()
 200
 201     if station['provider'] == 'python':
 202         LOG.debug("Getting custom data from external Python program")
 203     else:
 204         LOG.debug("Getting %s data from %s to %s",
 205                   str(station['desired_data']),
 206                   str(begin_date), str(end_date))
 207
 208     time_all_elements = time.time()
 209
 210     # get the data
 211     if station['provider'] == 'nrcs':
 212         infoex['wx_data'] = get_nrcs_data(begin_date, end_date, station)
 213     elif station['provider'] == 'mesowest':
 214         infoex['wx_data'] = get_mesowest_data(begin_date, end_date,
 215                                               station)
 216     elif station['provider'] == 'python':
 217         try:
 218             import importlib.util
 219
 220             spec = importlib.util.spec_from_file_location('custom_wx',
 221                                                           station['path'])
 222             mod = importlib.util.module_from_spec(spec)
 223             spec.loader.exec_module(mod)
 224             mod.LOG = LOG
 225
 226             try:
 227                 infoex['wx_data'] = mod.get_custom_data()
 228
 229                 if infoex['wx_data'] is None:
 230                     infoex['wx_data'] = []
 231             except Exception as exc:
 232                 LOG.error("Python program for custom Wx data failed in "
 233                           "execution: %s", str(exc))
 234                 sys.exit(1)
 235
 236             LOG.info("Successfully executed external Python program")
 237         except ImportError:
 238             LOG.error("Please upgrade to Python 3.3 or later")
 239             sys.exit(1)
 240         except FileNotFoundError:
 241             LOG.error("Specified Python program for custom Wx data "
 242                       "was not found")
 243             sys.exit(1)
 244         except Exception as exc:
 245             LOG.error("A problem was encountered when attempting to "
 246                       "load your custom Wx program: %s", str(exc))
 247             sys.exit(1)
 248
 249     LOG.info("Time taken to get all data : %.3f sec", time.time() -
 250              time_all_elements)
 251
 252     LOG.debug("infoex[wx_data]: %s", str(infoex['wx_data']))
 253
 254     # Now we only need to add in what we want to change thanks to that
 255     # abomination of a variable declaration earlier
 256     final_data[fmap['Location UUID']] = infoex['location_uuid']
 257     final_data[fmap['obDate']] = end_date.strftime('%m/%d/%Y')
 258     final_data[fmap['obTime']] = end_date.strftime('%H:%M')
 259
 260     for element_cd in infoex['wx_data']:
 261         if element_cd not in iemap:
 262             LOG.warning("BAD KEY wx_data['%s']", element_cd)
 263             continue
 264
 265         # Massage precision of certain values to fit InfoEx's
 266         # expectations
 267         #
 268         # 0 decimal places: wind speed, wind direction, wind gust, snow depth
 269         # 1 decimal place:  air temp, relative humidity, baro
 270         # Avoid transforming None values
 271         if infoex['wx_data'][element_cd] is None:
 272             continue
 273         elif element_cd in ['wind_speed', 'WSPD', 'wind_direction',
 274                             'WDIR', 'wind_gust', 'SNWD', 'snow_depth']:
 275             infoex['wx_data'][element_cd] = round(infoex['wx_data'][element_cd])
 276         elif element_cd in ['TOBS', 'air_temp', 'RHUM',
 277                             'relative_humidity', 'PRES', 'pressure']:
 278             infoex['wx_data'][element_cd] = round(infoex['wx_data'][element_cd], 1)
 279
 280         # CONSIDER: Casting every value to Float() -- need to investigate if
 281         #           any possible elementCds we may want are any other data
 282         #           type than float.
 283         #
 284         #           Another possibility is to query the API with
 285         #           getStationElements and temporarily store the
 286         #           storedUnitCd. But that's pretty network-intensive and
 287         #           may not even be worth it if there's only e.g. one or two
 288         #           exceptions to any otherwise uniformly Float value set.
 289         final_data[fmap[iemap[element_cd]]] = infoex['wx_data'][element_cd]
 290
 291     LOG.debug("final_data: %s", str(final_data))
 292
 293     if infoex['wx_data']:
 294         if not write_local_csv(infoex['csv_filename'], final_data):
 295             LOG.warning('Could not write local CSV file: %s',
 296                         infoex['csv_filename'])
 297             return 1
 298
 299         if not options.dry_run:
 300             upload_csv(infoex['csv_filename'], infoex)
 301
 302     LOG.debug('DONE')
 303     return 0
 304
 305 # data structure operations
 306 def setup_infoex_fields_mapping(location_uuid):
 307     """
 308     Create a mapping of InfoEx fields to the local data's indexing scheme.
 309
 310     INFOEX FIELDS
 311
 312     This won't earn style points in Python, but here we establish a couple
 313     of helpful mappings variables. The reason this is helpful is that the
 314     end result is simply an ordered set, the CSV file. But we still may
 315     want to manipulate the values arbitrarily before writing that file.
 316
 317     Also note that the current Auto Wx InfoEx documentation shows these
 318     keys in a graphical table with the "index" beginning at 1, but here we
 319     sanely index beginning at 0.
 320     """
 321     # pylint: disable=too-many-statements,multiple-statements,bad-whitespace
 322     fmap = {}                           ; final_data     = [None] * 29
 323     fmap['Location UUID'] = 0           ; final_data[0]  = location_uuid
 324     fmap['obDate'] = 1                  ; final_data[1]  = None
 325     fmap['obTime'] = 2                  ; final_data[2]  = None
 326     fmap['timeZone'] = 3                ; final_data[3]  = 'Pacific'
 327     fmap['tempMaxHour'] = 4             ; final_data[4]  = None
 328     fmap['tempMaxHourUnit'] = 5         ; final_data[5]  = 'F'
 329     fmap['tempMinHour'] = 6             ; final_data[6]  = None
 330     fmap['tempMinHourUnit'] = 7         ; final_data[7]  = 'F'
 331     fmap['tempPres'] = 8                ; final_data[8]  = None
 332     fmap['tempPresUnit'] = 9            ; final_data[9]  = 'F'
 333     fmap['precipitationGauge'] = 10     ; final_data[10] = None
 334     fmap['precipitationGaugeUnit'] = 11 ; final_data[11] = 'in'
 335     fmap['windSpeedNum'] = 12           ; final_data[12] = None
 336     fmap['windSpeedUnit'] = 13          ; final_data[13] = 'mph'
 337     fmap['windDirectionNum'] = 14       ; final_data[14] = None
 338     fmap['hS'] = 15                     ; final_data[15] = None
 339     fmap['hsUnit'] = 16                 ; final_data[16] = 'in'
 340     fmap['baro'] = 17                   ; final_data[17] = None
 341     fmap['baroUnit'] = 18               ; final_data[18] = 'inHg'
 342     fmap['rH'] = 19                     ; final_data[19] = None
 343     fmap['windGustSpeedNum'] = 20       ; final_data[20] = None
 344     fmap['windGustSpeedNumUnit'] = 21   ; final_data[21] = 'mph'
 345     fmap['windGustDirNum'] = 22         ; final_data[22] = None
 346     fmap['dewPoint'] = 23               ; final_data[23] = None
 347     fmap['dewPointUnit'] = 24           ; final_data[24] = 'F'
 348     fmap['hn24Auto'] = 25               ; final_data[25] = None
 349     fmap['hn24AutoUnit'] = 26           ; final_data[26] = 'in'
 350     fmap['hstAuto'] = 27                ; final_data[27] = None
 351     fmap['hstAutoUnit'] = 28            ; final_data[28] = 'in'
 352
 353     return (fmap, final_data)
 354
 355 def setup_infoex_counterparts_mapping(provider):
 356     """
 357     Create a mapping of the NRCS/MesoWest fields that this program supports to
 358     their InfoEx counterparts
 359     """
 360     iemap = {}
 361
 362     if provider == 'nrcs':
 363         iemap['PREC'] = 'precipitationGauge'
 364         iemap['TOBS'] = 'tempPres'
 365         iemap['TMAX'] = 'tempMaxHour'
 366         iemap['TMIN'] = 'tempMinHour'
 367         iemap['SNWD'] = 'hS'
 368         iemap['PRES'] = 'baro'
 369         iemap['RHUM'] = 'rH'
 370         iemap['WSPD'] = 'windSpeedNum'
 371         iemap['WDIR'] = 'windDirectionNum'
 372         # unsupported by NRCS:
 373         # windGustSpeedNum
 374     elif provider == 'mesowest':
 375         iemap['precip_accum'] = 'precipitationGauge'
 376         iemap['air_temp'] = 'tempPres'
 377         iemap['air_temp_high_24_hour'] = 'tempMaxHour'
 378         iemap['air_temp_low_24_hour'] = 'tempMinHour'
 379         iemap['snow_depth'] = 'hS'
 380         iemap['pressure'] = 'baro'
 381         iemap['relative_humidity'] = 'rH'
 382         iemap['wind_speed'] = 'windSpeedNum'
 383         iemap['wind_direction'] = 'windDirectionNum'
 384         iemap['wind_gust'] = 'windGustSpeedNum'
 385     elif provider == 'python':
 386         # we expect Python programs to use the InfoEx data type names
 387         iemap['precipitationGauge'] = 'precipitationGauge'
 388         iemap['tempPres'] = 'tempPres'
 389         iemap['tempMaxHour'] = 'tempMaxHour'
 390         iemap['tempMinHour'] = 'tempMinHour'
 391         iemap['hS'] = 'hS'
 392         iemap['baro'] = 'baro'
 393         iemap['rH'] = 'rH'
 394         iemap['windSpeedNum'] = 'windSpeedNum'
 395         iemap['windDirectionNum'] = 'windDirectionNum'
 396         iemap['windGustSpeedNum'] = 'windGustSpeedNum'
 397
 398     return iemap
 399
 400 # provider-specific operations
 401 def get_nrcs_data(begin, end, station):
 402     """get the data we're after from the NRCS WSDL"""
 403     transport = zeep.transports.Transport(cache=zeep.cache.SqliteCache())
 404     client = zeep.Client(wsdl=station['source'], transport=transport)
 405     remote_data = {}
 406
 407     for element_cd in station['desired_data']:
 408         time_element = time.time()
 409
 410         # get the last three hours of data for this elementCd/element_cd
 411         tmp = client.service.getHourlyData(
 412             stationTriplets=[station['station_id']],
 413             elementCd=element_cd,
 414             ordinal=1,
 415             beginDate=begin,
 416             endDate=end)
 417
 418         LOG.info("Time to get NRCS elementCd '%s': %.3f sec", element_cd,
 419                  time.time() - time_element)
 420
 421         values = tmp[0]['values']
 422
 423         # sort and isolate the most recent
 424         #
 425         # NOTE: we do this because sometimes there are gaps in hourly data
 426         #       in NRCS; yes, we may end up with slightly inaccurate data,
 427         #       so perhaps this decision will be re-evaluated in the future
 428         if values:
 429             ordered = sorted(values, key=lambda t: t['dateTime'], reverse=True)
 430             remote_data[element_cd] = ordered[0]['value']
 431         else:
 432             remote_data[element_cd] = None
 433
 434     return remote_data
 435
 436 def get_mesowest_data(begin, end, station):
 437     """get the data we're after from the MesoWest/Synoptic API"""
 438     remote_data = {}
 439
 440     # massage begin/end date format
 441     begin_date_str = begin.strftime('%Y%m%d%H%M')
 442     end_date_str = end.strftime('%Y%m%d%H%M')
 443
 444     # construct final, completed API URL
 445     api_req_url = station['source'] + '&start=' + begin_date_str + '&end=' + end_date_str
 446     req = requests.get(api_req_url)
 447
 448     try:
 449         json = req.json()
 450     except ValueError:
 451         LOG.error("Bad JSON in MesoWest response")
 452         sys.exit(1)
 453
 454     try:
 455         observations = json['STATION'][0]['OBSERVATIONS']
 456     except ValueError:
 457         LOG.error("Bad JSON in MesoWest response")
 458         sys.exit(1)
 459
 460     pos = len(observations['date_time']) - 1
 461
 462     for element_cd in station['desired_data'].split(','):
 463         # sort and isolate the most recent, see note above in NRCS for how and
 464         # why this is done
 465         #
 466         # NOTE: Unlike in the NRCS case, the MesoWest API response contains all
 467         #       data (whereas with NRCS, we have to make a separate request for
 468         #       each element we want). This is nice for network efficiency but
 469         #       it means we have to handle this part differently for each.
 470         #
 471         # NOTE: Also unlike NRCS, MesoWest provides more granular data; NRCS
 472         #       provides hourly data, but MesoWest can often provide data every
 473         #       10 minutes -- though this provides more opportunity for
 474         #       irregularities
 475
 476         # we may not have the data at all
 477         key_name = element_cd + '_set_1'
 478
 479         if key_name in observations:
 480             if observations[key_name][pos]:
 481                 remote_data[element_cd] = observations[key_name][pos]
 482
 483                 # mesowest provides wind_speed in m/s, we want mph
 484                 if element_cd in ('wind_speed', 'wind_gust'):
 485                     remote_data[element_cd] = ms_to_mph(remote_data[element_cd])
 486             else:
 487                 remote_data[element_cd] = None
 488         else:
 489             remote_data[element_cd] = None
 490
 491     return remote_data
 492
 493 def switch_units_to_metric(data_map, mapping):
 494     """replace units with metric counterparts"""
 495
 496     # NOTE: to update this, use the fmap<->final_data mapping laid out
 497     #       in setup_infoex_fields_mapping ()
 498     #
 499     # NOTE: this only 'works' with MesoWest for now, as the MesoWest API
 500     #       itself handles the unit conversion; in the future, we will also
 501     #       support NRCS unit conversion, but this must be done by this
 502     #       program.
 503     data_map[mapping['tempPresUnit']] = 'C'
 504     data_map[mapping['hsUnit']] = 'm'
 505     data_map[mapping['windSpeedUnit']] = 'm/s'
 506     data_map[mapping['windGustSpeedNumUnit']] = 'm/s'
 507
 508     return data_map
 509
 510 # CSV operations
 511 def write_local_csv(path_to_file, data):
 512     """Write the specified CSV file to disk"""
 513     with open(path_to_file, 'w') as file_object:
 514         # The requirement is that empty values are represented in the CSV
 515         # file as "", csv.QUOTE_NONNUMERIC achieves that
 516         LOG.debug("writing CSV file '%s'", path_to_file)
 517         writer = csv.writer(file_object, quoting=csv.QUOTE_NONNUMERIC)
 518         writer.writerow(data)
 519         file_object.close()
 520     return True
 521
 522 def upload_csv(path_to_file, infoex_data):
 523     """Upload the specified CSV file to InfoEx FTP and remove the file"""
 524     with open(path_to_file, 'rb') as file_object:
 525         LOG.debug("uploading FTP file '%s'", infoex_data['host'])
 526         ftp = FTP(infoex_data['host'], infoex_data['uuid'],
 527                   infoex_data['api_key'])
 528         ftp.storlines('STOR ' + path_to_file, file_object)
 529         ftp.close()
 530         file_object.close()
 531     os.remove(path_to_file)
 532
 533 # other miscellaneous routines
 534 def setup_time_values():
 535     """establish time bounds of data request(s)"""
 536     # floor time to nearest hour
 537     date_time = datetime.datetime.now()
 538     end_date = date_time - datetime.timedelta(minutes=date_time.minute % 60,
 539                                               seconds=date_time.second,
 540                                               microseconds=date_time.microsecond)
 541     begin_date = end_date - datetime.timedelta(hours=3)
 542     return (begin_date, end_date)
 543
 544 def ms_to_mph(ms):
 545     """convert meters per second to miles per hour"""
 546     return ms * 2.236936
 547
 548 if __name__ == "__main__":
 549     sys.exit(main())