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__ = '2.2.0'
  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:
 232                 LOG.error("Python program for custom Wx data failed in "
 233                           "execution")
 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
 245     LOG.info("Time taken to get all data : %.3f sec", time.time() -
 246              time_all_elements)
 247
 248     LOG.debug("infoex[wx_data]: %s", str(infoex['wx_data']))
 249
 250     # Now we only need to add in what we want to change thanks to that
 251     # abomination of a variable declaration earlier
 252     final_data[fmap['Location UUID']] = infoex['location_uuid']
 253     final_data[fmap['obDate']] = end_date.strftime('%m/%d/%Y')
 254     final_data[fmap['obTime']] = end_date.strftime('%H:%M')
 255
 256     for element_cd in infoex['wx_data']:
 257         if element_cd not in iemap:
 258             LOG.warning("BAD KEY wx_data['%s']", element_cd)
 259             continue
 260
 261         # Massage precision of certain values to fit InfoEx's
 262         # expectations
 263         #
 264         # 0 decimal places: wind speed, wind direction, wind gust, snow depth
 265         # 1 decimal place:  air temp, baro
 266         # Avoid transforming None values
 267         if infoex['wx_data'][element_cd] is None:
 268             continue
 269         elif element_cd in ['wind_speed', 'WSPD', 'wind_direction',
 270                             'WDIR', 'wind_gust', 'SNWD', 'snow_depth']:
 271             infoex['wx_data'][element_cd] = round(infoex['wx_data'][element_cd])
 272         elif element_cd in ['TOBS', 'air_temp', 'PRES', 'pressure']:
 273             infoex['wx_data'][element_cd] = round(infoex['wx_data'][element_cd], 1)
 274
 275         # CONSIDER: Casting every value to Float() -- need to investigate if
 276         #           any possible elementCds we may want are any other data
 277         #           type than float.
 278         #
 279         #           Another possibility is to query the API with
 280         #           getStationElements and temporarily store the
 281         #           storedUnitCd. But that's pretty network-intensive and
 282         #           may not even be worth it if there's only e.g. one or two
 283         #           exceptions to any otherwise uniformly Float value set.
 284         final_data[fmap[iemap[element_cd]]] = infoex['wx_data'][element_cd]
 285
 286     LOG.debug("final_data: %s", str(final_data))
 287
 288     if len(infoex['wx_data']) > 0:
 289         if not write_local_csv(infoex['csv_filename'], final_data):
 290             LOG.warning('Could not write local CSV file: %s',
 291                         infoex['csv_filename'])
 292             return 1
 293
 294         if not options.dry_run:
 295             upload_csv(infoex['csv_filename'], infoex)
 296
 297     LOG.debug('DONE')
 298     return 0
 299
 300 # data structure operations
 301 def setup_infoex_fields_mapping(location_uuid):
 302     """
 303     Create a mapping of InfoEx fields to the local data's indexing scheme.
 304
 305     INFOEX FIELDS
 306
 307     This won't earn style points in Python, but here we establish a couple
 308     of helpful mappings variables. The reason this is helpful is that the
 309     end result is simply an ordered set, the CSV file. But we still may
 310     want to manipulate the values arbitrarily before writing that file.
 311
 312     Also note that the current Auto Wx InfoEx documentation shows these
 313     keys in a graphical table with the "index" beginning at 1, but here we
 314     sanely index beginning at 0.
 315     """
 316     # pylint: disable=too-many-statements,multiple-statements,bad-whitespace
 317     fmap = {}                           ; final_data     = [None] * 29
 318     fmap['Location UUID'] = 0           ; final_data[0]  = location_uuid
 319     fmap['obDate'] = 1                  ; final_data[1]  = None
 320     fmap['obTime'] = 2                  ; final_data[2]  = None
 321     fmap['timeZone'] = 3                ; final_data[3]  = 'Pacific'
 322     fmap['tempMaxHour'] = 4             ; final_data[4]  = None
 323     fmap['tempMaxHourUnit'] = 5         ; final_data[5]  = 'F'
 324     fmap['tempMinHour'] = 6             ; final_data[6]  = None
 325     fmap['tempMinHourUnit'] = 7         ; final_data[7]  = 'F'
 326     fmap['tempPres'] = 8                ; final_data[8]  = None
 327     fmap['tempPresUnit'] = 9            ; final_data[9]  = 'F'
 328     fmap['precipitationGauge'] = 10     ; final_data[10] = None
 329     fmap['precipitationGaugeUnit'] = 11 ; final_data[11] = 'in'
 330     fmap['windSpeedNum'] = 12           ; final_data[12] = None
 331     fmap['windSpeedUnit'] = 13          ; final_data[13] = 'mph'
 332     fmap['windDirectionNum'] = 14       ; final_data[14] = None
 333     fmap['hS'] = 15                     ; final_data[15] = None
 334     fmap['hsUnit'] = 16                 ; final_data[16] = 'in'
 335     fmap['baro'] = 17                   ; final_data[17] = None
 336     fmap['baroUnit'] = 18               ; final_data[18] = 'inHg'
 337     fmap['rH'] = 19                     ; final_data[19] = None
 338     fmap['windGustSpeedNum'] = 20       ; final_data[20] = None
 339     fmap['windGustSpeedNumUnit'] = 21   ; final_data[21] = 'mph'
 340     fmap['windGustDirNum'] = 22         ; final_data[22] = None
 341     fmap['dewPoint'] = 23               ; final_data[23] = None
 342     fmap['dewPointUnit'] = 24           ; final_data[24] = 'F'
 343     fmap['hn24Auto'] = 25               ; final_data[25] = None
 344     fmap['hn24AutoUnit'] = 26           ; final_data[26] = 'in'
 345     fmap['hstAuto'] = 27                ; final_data[27] = None
 346     fmap['hstAutoUnit'] = 28            ; final_data[28] = 'in'
 347
 348     return (fmap, final_data)
 349
 350 def setup_infoex_counterparts_mapping(provider):
 351     """
 352     Create a mapping of the NRCS/MesoWest fields that this program supports to
 353     their InfoEx counterparts
 354     """
 355     iemap = {}
 356
 357     if provider == 'nrcs':
 358         iemap['PREC'] = 'precipitationGauge'
 359         iemap['TOBS'] = 'tempPres'
 360         iemap['TMAX'] = 'tempMaxHour'
 361         iemap['TMIN'] = 'tempMinHour'
 362         iemap['SNWD'] = 'hS'
 363         iemap['PRES'] = 'baro'
 364         iemap['RHUM'] = 'rH'
 365         iemap['WSPD'] = 'windSpeedNum'
 366         iemap['WDIR'] = 'windDirectionNum'
 367         # unsupported by NRCS:
 368         # windGustSpeedNum
 369     elif provider == 'mesowest':
 370         iemap['precip_accum'] = 'precipitationGauge'
 371         iemap['air_temp'] = 'tempPres'
 372         iemap['air_temp_high_24_hour'] = 'tempMaxHour'
 373         iemap['air_temp_low_24_hour'] = 'tempMinHour'
 374         iemap['snow_depth'] = 'hS'
 375         iemap['pressure'] = 'baro'
 376         iemap['relative_humidity'] = 'rH'
 377         iemap['wind_speed'] = 'windSpeedNum'
 378         iemap['wind_direction'] = 'windDirectionNum'
 379         iemap['wind_gust'] = 'windGustSpeedNum'
 380     elif provider == 'python':
 381         # we expect Python programs to use the InfoEx data type names
 382         iemap['precipitationGauge'] = 'precipitationGauge'
 383         iemap['tempPres'] = 'tempPres'
 384         iemap['tempMaxHour'] = 'tempMaxHour'
 385         iemap['tempMinHour'] = 'tempMinHour'
 386         iemap['hS'] = 'hS'
 387         iemap['baro'] = 'baro'
 388         iemap['rH'] = 'rH'
 389         iemap['windSpeedNum'] = 'windSpeedNum'
 390         iemap['windDirectionNum'] = 'windDirectionNum'
 391         iemap['windGustSpeedNum'] = 'windGustSpeedNum'
 392
 393     return iemap
 394
 395 # provider-specific operations
 396 def get_nrcs_data(begin, end, station):
 397     """get the data we're after from the NRCS WSDL"""
 398     transport = zeep.transports.Transport(cache=zeep.cache.SqliteCache())
 399     client = zeep.Client(wsdl=station['source'], transport=transport)
 400     remote_data = {}
 401
 402     for element_cd in station['desired_data']:
 403         time_element = time.time()
 404
 405         # get the last three hours of data for this elementCd/element_cd
 406         tmp = client.service.getHourlyData(
 407             stationTriplets=[station['station_id']],
 408             elementCd=element_cd,
 409             ordinal=1,
 410             beginDate=begin,
 411             endDate=end)
 412
 413         LOG.info("Time to get NRCS elementCd '%s': %.3f sec", element_cd,
 414                  time.time() - time_element)
 415
 416         values = tmp[0]['values']
 417
 418         # sort and isolate the most recent
 419         #
 420         # NOTE: we do this because sometimes there are gaps in hourly data
 421         #       in NRCS; yes, we may end up with slightly inaccurate data,
 422         #       so perhaps this decision will be re-evaluated in the future
 423         if values:
 424             ordered = sorted(values, key=lambda t: t['dateTime'], reverse=True)
 425             remote_data[element_cd] = ordered[0]['value']
 426         else:
 427             remote_data[element_cd] = None
 428
 429     return remote_data
 430
 431 def get_mesowest_data(begin, end, station):
 432     """get the data we're after from the MesoWest/Synoptic API"""
 433     remote_data = {}
 434
 435     # massage begin/end date format
 436     begin_date_str = begin.strftime('%Y%m%d%H%M')
 437     end_date_str = end.strftime('%Y%m%d%H%M')
 438
 439     # construct final, completed API URL
 440     api_req_url = station['source'] + '&start=' + begin_date_str + '&end=' + end_date_str
 441     req = requests.get(api_req_url)
 442
 443     try:
 444         json = req.json()
 445     except ValueError:
 446         LOG.error("Bad JSON in MesoWest response")
 447         sys.exit(1)
 448
 449     try:
 450         observations = json['STATION'][0]['OBSERVATIONS']
 451     except ValueError:
 452         LOG.error("Bad JSON in MesoWest response")
 453         sys.exit(1)
 454
 455     pos = len(observations['date_time']) - 1
 456
 457     for element_cd in station['desired_data'].split(','):
 458         # sort and isolate the most recent, see note above in NRCS for how and
 459         # why this is done
 460         #
 461         # NOTE: Unlike in the NRCS case, the MesoWest API response contains all
 462         #       data (whereas with NRCS, we have to make a separate request for
 463         #       each element we want). This is nice for network efficiency but
 464         #       it means we have to handle this part differently for each.
 465         #
 466         # NOTE: Also unlike NRCS, MesoWest provides more granular data; NRCS
 467         #       provides hourly data, but MesoWest can often provide data every
 468         #       10 minutes -- though this provides more opportunity for
 469         #       irregularities
 470
 471         # we may not have the data at all
 472         key_name = element_cd + '_set_1'
 473         if key_name in observations:
 474             if observations[key_name][pos]:
 475                 remote_data[element_cd] = observations[key_name][pos]
 476             else:
 477                 remote_data[element_cd] = None
 478         else:
 479             remote_data[element_cd] = None
 480
 481     return remote_data
 482
 483 def switch_units_to_metric(data_map, mapping):
 484     """replace units with metric counterparts"""
 485
 486     # NOTE: to update this, use the fmap<->final_data mapping laid out
 487     #       in setup_infoex_fields_mapping ()
 488     #
 489     # NOTE: this only 'works' with MesoWest for now, as the MesoWest API
 490     #       itself handles the unit conversion; in the future, we will also
 491     #       support NRCS unit conversion, but this must be done by this
 492     #       program.
 493     data_map[mapping['tempPresUnit']] = 'C'
 494     data_map[mapping['hsUnit']] = 'm'
 495     data_map[mapping['windSpeedUnit']] = 'm/s'
 496     data_map[mapping['windGustSpeedNumUnit']] = 'm/s'
 497
 498     return data_map
 499
 500 # CSV operations
 501 def write_local_csv(path_to_file, data):
 502     """Write the specified CSV file to disk"""
 503     with open(path_to_file, 'w') as file_object:
 504         # The requirement is that empty values are represented in the CSV
 505         # file as "", csv.QUOTE_NONNUMERIC achieves that
 506         LOG.debug("writing CSV file '%s'", path_to_file)
 507         writer = csv.writer(file_object, quoting=csv.QUOTE_NONNUMERIC)
 508         writer.writerow(data)
 509         file_object.close()
 510     return True
 511
 512 def upload_csv(path_to_file, infoex_data):
 513     """Upload the specified CSV file to InfoEx FTP and remove the file"""
 514     with open(path_to_file, 'rb') as file_object:
 515         LOG.debug("uploading FTP file '%s'", infoex_data['host'])
 516         ftp = FTP(infoex_data['host'], infoex_data['uuid'],
 517                   infoex_data['api_key'])
 518         ftp.storlines('STOR ' + path_to_file, file_object)
 519         ftp.close()
 520         file_object.close()
 521     os.remove(path_to_file)
 522
 523 # other miscellaneous routines
 524 def setup_time_values():
 525     """establish time bounds of data request(s)"""
 526     # floor time to nearest hour
 527     date_time = datetime.datetime.now()
 528     end_date = date_time - datetime.timedelta(minutes=date_time.minute % 60,
 529                                               seconds=date_time.second,
 530                                               microseconds=date_time.microsecond)
 531     begin_date = end_date - datetime.timedelta(hours=3)
 532     return (begin_date, end_date)
 533
 534 if __name__ == "__main__":
 535     sys.exit(main())