def build_list_dates_generic(from_date, to_date, frequency_id):
# Add a check on frequency
try:
frequency = datasets.Dataset.get_frequency(
frequency_id, datasets.Frequency.DATEFORMAT.DATETIME)
except Exception as inst:
logger.debug("Error in datasets.Dataset.get_frequency: %s" %
inst.args[0])
raise
# Manage the start_date (mandatory).
try:
# If it is a date, convert to datetime
if functions.is_date_yyyymmdd(str(from_date), silent=True):
datetime_start = datetime.datetime.strptime(
str(from_date), '%Y%m%d')
else:
# If it is a negative number, subtract from current date
if isinstance(from_date, int) or isinstance(from_date, int):
if from_date < 0:
datetime_start = datetime.datetime.today(
) - datetime.timedelta(days=-from_date)
else:
logger.debug("Error in Start Date: must be YYYYMMDD or -Ndays")
raise Exception("Start Date not valid")
except:
raise Exception("Start Date not valid")
# Manage the end_date (mandatory).
try:
if functions.is_date_yyyymmdd(str(to_date), silent=True):
datetime_end = datetime.datetime.strptime(str(to_date), '%Y%m%d')
# If it is a negative number, subtract from current date
elif isinstance(to_date, int) or isinstance(to_date, int):
if to_date < 0:
datetime_end = datetime.datetime.today() - datetime.timedelta(
days=-to_date)
elif to_date > 0:
datetime_end = datetime.datetime.today() + datetime.timedelta(
days=to_date)
else:
datetime_end = datetime.datetime.today()
except:
pass
try:
dates = frequency.get_dates(datetime_start, datetime_end)
except Exception as inst:
logger.debug("Error in frequency.get_dates: %s" % inst.args[0])
raise
return dates
def syncGeoserver():
#
# Copy some 'relevant' datasets to GeoServer
# Selection of datasets is done on the basis of the product.geoserver table
#
# Get list of all 'relevant' subproducts (see 2. above)
list_active_geoserver = esTools.get_activated_geoserver()
# Loop over existing sub_products
for geoserver_sprod in list_active_geoserver:
# Extract local variable:
my_prod = geoserver_sprod.productcode
my_subprod = geoserver_sprod.subproductcode
my_version = geoserver_sprod.version
start_date = geoserver_sprod.startdate
end_date = geoserver_sprod.enddate
logger.info("Working on Product/Subproduct/Version: {0}/{1}/{2}".format(my_prod, my_subprod, my_version))
# Manage dates from bigint to datetime
if functions.is_date_yyyymmdd(str(start_date), silent=True):
date_start = datetime.datetime.strptime(str(start_date), '%Y%m%d').date()
else:
date_start = None
if functions.is_date_yyyymmdd(str(end_date), silent=True):
date_end = datetime.datetime.strptime(str(end_date), '%Y%m%d').date()
else:
date_end = None
# Get additional products info
product_info = querydb.get_product_out_info(productcode=my_prod,
subproductcode=my_subprod,
version=my_version)
# my_mapset = subprod.mapsetcode
my_type = product_info[0].product_type
my_category = product_info[0].category_id
# Create a Product object (to get mapsets)
my_product = products.Product(my_prod, version=my_version)
my_mapsets = my_product.mapsets
if len(my_mapsets) > 1:
logger.info('More than 1 mapset exists. Take the first')
if len(my_mapsets) == 0:
logger.warning('No any mapset exists. Skip.')
continue
my_mapset = my_mapsets[0]
# Create a Dataset object (to get file list)
# If data_start is not set (e.g. for 10davg prod) create w/o dates
if date_start:
my_dataset = datasets.Dataset(my_prod, my_subprod, my_mapset, version=my_version, from_date=date_start,
to_date=date_end)
if my_dataset._frequency.dateformat == 'MMDD':
logger.warning('Product of type MMDD: date specification not supported. Skip.')
continue
file_list = my_dataset.get_filenames_range()
else:
my_dataset = datasets.Dataset(my_prod, my_subprod, my_mapset, version=my_version)
file_list = my_dataset.get_filenames()
# Check that there is at least 1 file
if len(file_list) > 0:
# Check the Workspace exists, or create it
my_workspace = esTools.setWorkspaceName(my_category, my_prod, my_subprod, my_version, my_mapset,
nameType=geoserverREST.geoserverWorkspaceName)
if not geoserverREST.isWorkspace(my_workspace):
geoserverREST.createWorkspace(my_workspace)
# Loop over files and upload
for my_file in file_list:
my_date = functions.get_date_from_path_full(my_file)
# if subprod in list_active_subprods:
logger.debug("Working on Product/Subproduct/Version/Mapset/Date: {0}/{1}/{2}/{3}/{4}".format(
my_prod, my_subprod, my_version, my_mapset, my_date))
# Upload the file and register
esTools.uploadAndRegisterRaster(my_category, my_prod, my_subprod, my_version, my_mapset, my_date,
my_type, local_data_dir)
def loop_get_internet(dry_run=False):
global processed_list_filename, processed_list
global processed_info_filename, processed_info
signal.signal(signal.SIGTERM, signal_handler)
signal.signal(signal.SIGINT, signal_handler)
signal.signal(signal.SIGILL, signal_handler)
logger.info("Starting retrieving data from INTERNET.")
while True:
output_dir = es_constants.ingest_dir
logger.debug("Check if the Ingest Server input directory : %s exists.", output_dir)
if not os.path.exists(output_dir):
logger.fatal("The Ingest Server input directory : %s doesn't exists.", output_dir)
exit(1)
if not os.path.exists(es_constants.processed_list_int_dir):
os.mkdir(es_constants.processed_list_int_dir)
while 1:
try:
time_sleep = user_def_sleep
logger.debug("Sleep time set to : %s.", time_sleep)
except:
logger.warning("Sleep time not defined. Setting to default=1min. Continue.")
time_sleep = 60
logger.debug("Reading active INTERNET data sources from database")
internet_sources_list = querydb.get_active_internet_sources(echo=echo_query)
# Loop over active triggers
try:
for internet_source in internet_sources_list:
logger.debug("Processing internet source %s.", internet_source.descriptive_name)
processed_list_filename = es_constants.get_internet_processed_list_prefix+str(internet_source.internet_id)+'.list'
processed_info_filename = es_constants.get_internet_processed_list_prefix+str(internet_source.internet_id)+'.info'
# Create objects for list and info
processed_list = []
processed_info = {'length_proc_list': 0,
'time_latest_exec': datetime.datetime.now(),
'time_latest_copy': datetime.datetime.now()}
# Restore/Create List
processed_list=functions.restore_obj_from_pickle(processed_list, processed_list_filename)
# Restore/Create Info
processed_info=functions.restore_obj_from_pickle(processed_info, processed_info_filename)
# Update processing time (in case it is restored)
processed_info['time_latest_exec']=datetime.datetime.now()
logger.debug("Create current list of file to process for source %s.", internet_source.internet_id)
if internet_source.user_name is None:
user_name = "anonymous"
else:
user_name = internet_source.user_name
if internet_source.password is None:
password = "******"
else:
password = internet_source.password
usr_pwd = str(user_name)+':'+str(password)
logger.debug(" Url is %s.", internet_source.url)
logger.debug(" usr/pwd is %s.", usr_pwd)
logger.debug(" regex is %s.", internet_source.include_files_expression)
internet_type = internet_source.type
if internet_type == 'ftp':
# Note that the following list might contain sub-dirs (it reflects full_regex)
current_list = get_list_matching_files_dir_ftp(str(internet_source.url), str(usr_pwd), str(internet_source.include_files_expression))
elif internet_type == 'http_tmpl':
# Manage the dates:start_date is mandatory .. end_date replaced by 'today' if missing/wrong
try:
if functions.is_date_yyyymmdd(str(internet_source.start_date), silent=True):
datetime_start=datetime.datetime.strptime(str(internet_source.start_date),'%Y%m%d')
else:
raise Exception("Start Date not valid")
except:
raise Exception("Start Date not valid")
try:
if functions.is_date_yyyymmdd(str(internet_source.end_date), silent=True):
datetime_end=datetime.datetime.strptime(str(internet_source.end_date),'%Y%m%d')
else:
datetime_end=datetime.datetime.today()
except:
pass
# Create the full filename from a 'template' which contains
try:
current_list = build_list_matching_for_http(str(internet_source.url),
str(internet_source.include_files_expression),
datetime_start,
datetime_end,
str(internet_source.frequency_id))
except:
logger.error("Error in creating date lists. Continue")
logger.debug("Number of files currently available for source %s is %i", internet_source.internet_id, len(current_list))
if len(current_list) > 0:
logger.debug("Number of files already copied for trigger %s is %i", internet_source.internet_id, len(processed_list))
listtoprocess = []
for current_file in current_list:
if len(processed_list) == 0:
listtoprocess.append(current_file)
else:
#if os.path.basename(current_file) not in processed_list: -> save in .list subdirs as well !!
if current_file not in processed_list:
listtoprocess.append(current_file)
logger.debug("Number of files to be copied for trigger %s is %i", internet_source.internet_id, len(listtoprocess))
if listtoprocess != set([]):
logger.debug("Loop on the found files.")
if not dry_run:
for filename in list(listtoprocess):
logger.debug("Processing file: "+str(internet_source.url)+os.path.sep+filename)
try:
result = get_file_from_url(str(internet_source.url)+os.path.sep+filename, target_file=os.path.basename(filename), target_dir=es_constants.ingest_dir, userpwd=str(usr_pwd))
if not result:
logger.info("File %s copied.", filename)
processed_list.append(filename)
except:
logger.warning("Problem while copying file: %s.", filename)
else:
logger.info('Dry_run is set: do not get files')
if not dry_run:
functions.dump_obj_to_pickle(processed_list, processed_list_filename)
functions.dump_obj_to_pickle(processed_info, processed_info_filename)
sleep(float(user_def_sleep))
# Loop over sources
except Exception as inst:
logger.error("Error while processing source %s. Continue" % internet_source.descriptive_name)
sleep(float(user_def_sleep))
exit(0)
def define_output_data_format(datasource_descr, in_date, out_date_format):
# Convert the in_date format into a convenient one for DB and file naming
# (i.e YYYYMMDD or YYYYMMDDHHMM)
# Initialize to error value
output_date_str = -1
if datasource_descr.date_format == 'YYYYMMDD':
if functions.is_date_yyyymmdd(in_date):
output_date_str = in_date
else:
output_date_str = -1
if datasource_descr.date_format == 'YYYYMMDDHHMM':
if functions.is_date_yyyymmddhhmm(in_date):
output_date_str = in_date
else:
output_date_str = -1
if datasource_descr.date_format == 'YYYYDOY_YYYYDOY':
output_date_str = functions.conv_date_yyyydoy_2_yyyymmdd(
str(in_date)[0:7])
if datasource_descr.date_format == 'YYYYMMDD_YYYYMMDD':
output_date_str = str(in_date)[0:8]
if not functions.is_date_yyyymmdd(output_date_str):
output_date_str = -1
if datasource_descr.date_format == 'YYYYDOY':
output_date_str = functions.conv_date_yyyydoy_2_yyyymmdd(in_date)
if datasource_descr.date_format == 'YYYY_MM_DKX':
output_date_str = functions.conv_yyyy_mm_dkx_2_yyyymmdd(in_date)
if datasource_descr.date_format == 'YYMMK':
output_date_str = functions.conv_yymmk_2_yyyymmdd(in_date)
if datasource_descr.date_format == 'YYYYdMMdK':
output_date_str = functions.conv_yyyydmmdk_2_yyyymmdd(in_date)
if datasource_descr.date_format == 'YYYYMMDD_G2':
# The date (e.g. 20151103) is converted to the dekad it belongs to (e.g. 20151101)
output_date_str = functions.conv_yyyymmdd_g2_2_yyyymmdd(in_date)
if datasource_descr.date_format == 'MMDD':
output_date_str = str(in_date)
if datasource_descr.date_format == 'YYYYMM':
# Convert from YYYYMM -> YYYYMMDD
output_date_str = str(in_date) + '01'
if datasource_descr.date_format == 'YYYY_DK':
# The date (e.g. 2020_36) is converted to the dekad it belongs to (e.g. 20201221)
output_date_str = functions.conv_yyyydk_2_yyyymmdd(in_date)
if output_date_str == -1:
out_date_str_final = in_date + '_DATE_ERROR_'
else:
if out_date_format == 'YYYYMMDDHHMM':
if functions.is_date_yyyymmddhhmm(output_date_str):
out_date_str_final = output_date_str
elif functions.is_date_yyyymmdd(output_date_str):
out_date_str_final = output_date_str + '0000'
elif out_date_format == 'YYYYMMDD':
if functions.is_date_yyyymmdd(output_date_str, silent=True):
out_date_str_final = output_date_str
elif functions.is_date_yyyymmddhhmm(output_date_str):
out_date_str_final = output_date_str[0:8]
elif out_date_format == 'MMDD':
if functions.is_date_mmdd(output_date_str):
out_date_str_final = output_date_str
return out_date_str_final
def build_list_matching_files_motu(base_url, template, from_date, to_date,
frequency_id, username, password,
files_filter_expression):
# Add a check on frequency
try:
frequency = datasets.Dataset.get_frequency(
frequency_id, datasets.Frequency.DATEFORMAT.DATETIME)
except Exception as inst:
logger.debug("Error in datasets.Dataset.get_frequency: %s" %
inst.args[0])
raise
# Manage the start_date (mandatory).
try:
# If it is a date, convert to datetime
if functions.is_date_yyyymmdd(str(from_date), silent=True):
datetime_start = datetime.datetime.strptime(
str(from_date), '%Y%m%d')
else:
# If it is a negative number, subtract from current date
if isinstance(from_date, int) or isinstance(from_date, long):
if from_date < 0:
datetime_start = datetime.datetime.today(
) - datetime.timedelta(days=-from_date)
else:
logger.debug("Error in Start Date: must be YYYYMMDD or -Ndays")
raise Exception("Start Date not valid")
except:
raise Exception("Start Date not valid")
# Manage the end_date (mandatory).
try:
if functions.is_date_yyyymmdd(str(to_date), silent=True):
datetime_end = datetime.datetime.strptime(str(to_date), '%Y%m%d')
# If it is a negative number, subtract from current date
elif isinstance(to_date, int) or isinstance(to_date, long):
if to_date < 0:
datetime_end = datetime.datetime.today() - datetime.timedelta(
days=-to_date)
elif to_date > 0:
datetime_end = datetime.datetime.today() + datetime.timedelta(
days=to_date)
else:
datetime_end = datetime.datetime.today()
except:
pass
try:
dates = frequency.get_dates(datetime_start, datetime_end)
except Exception as inst:
logger.debug("Error in frequency.get_dates: %s" % inst.args[0])
raise
try:
list_filenames = motu_api.motu_4_dates(dates, template, base_url,
username, password,
files_filter_expression)
#list_filenames = frequency.get_internet_dates(dates, template)
except Exception as inst:
logger.debug("Error in motu_api.motu_getlists: %s" % inst.args[0])
raise
return list_filenames
def build_list_matching_files_tmpl(base_url, template, from_date, to_date,
frequency_id):
# Add a check on frequency
try:
frequency = datasets.Dataset.get_frequency(
frequency_id, datasets.Frequency.DATEFORMAT.DATETIME)
except Exception as inst:
logger.debug("Error in datasets.Dataset.get_frequency: %s" %
inst.args[0])
raise
# Manage the start_date (mandatory).
try:
# If it is a date, convert to datetime
if functions.is_date_yyyymmdd(str(from_date), silent=True):
datetime_start = datetime.datetime.strptime(
str(from_date), '%Y%m%d')
else:
# If it is a negative number, subtract from current date
if isinstance(from_date, int) or isinstance(from_date, long):
if from_date < 0:
datetime_start = datetime.datetime.today(
) - datetime.timedelta(days=-from_date)
else:
logger.debug("Error in Start Date: must be YYYYMMDD or -Ndays")
raise Exception("Start Date not valid")
except:
raise Exception("Start Date not valid")
# Manage the end_date (mandatory).
try:
if functions.is_date_yyyymmdd(str(to_date), silent=True):
datetime_end = datetime.datetime.strptime(str(to_date), '%Y%m%d')
# If it is a negative number, subtract from current date
elif isinstance(to_date, int) or isinstance(to_date, long):
if to_date < 0:
datetime_end = datetime.datetime.today() - datetime.timedelta(
days=-to_date)
else:
datetime_end = datetime.datetime.today()
except:
pass
try:
dates = frequency.get_dates(datetime_start, datetime_end)
except Exception as inst:
logger.debug("Error in frequency.get_dates: %s" % inst.args[0])
raise
try:
if sys.platform == 'win32':
template.replace("-", "#")
list_filenames = frequency.get_internet_dates(dates, template)
except Exception as inst:
logger.debug("Error in frequency.get_internet_dates: %s" %
inst.args[0])
raise
return list_filenames
######################################################################################
# build_list_matching_files_sentinel_sat
# Purpose: return the list of file names matching a 'template' with 'date' placeholders
# It is the entry point for the 'http_templ' source type
# Author: Vijay Charan Venkatachalam, JRC, European Commission
# Date: 2015/02/18
# Inputs: template: regex including subdirs (e.g. 'Collection51/TIFF/Win1[01]/201[1-3]/MCD45monthly.A20.*burndate.tif.gz'
# from_date: start date for the dataset (datetime.datetime object)
# to_date: end date for the dataset (datetime.datetime object)
# frequency: dataset 'frequency' (see DB 'frequency' table)
#
# def build_list_matching_files_sentinel_sat(base_url, template, from_date, to_date, frequency_id, username, password):
#
# # Add a check on frequency
# try:
# frequency = datasets.Dataset.get_frequency(frequency_id, datasets.Frequency.DATEFORMAT.DATETIME)
# except Exception as inst:
# logger.debug("Error in datasets.Dataset.get_frequency: %s" %inst.args[0])
# raise
#
# # Manage the start_date (mandatory).
# try:
# # If it is a date, convert to datetime
# if functions.is_date_yyyymmdd(str(from_date), silent=True):
# datetime_start=datetime.datetime.strptime(str(from_date),'%Y%m%d')
# else:
# # If it is a negative number, subtract from current date
# if isinstance(from_date,int) or isinstance(from_date,long):
# if from_date < 0:
# datetime_start=datetime.datetime.today() - datetime.timedelta(days=-from_date)
# else:
# logger.debug("Error in Start Date: must be YYYYMMDD or -Ndays")
# raise Exception("Start Date not valid")
# except:
# raise Exception("Start Date not valid")
#
# # Manage the end_date (mandatory).
# try:
# if functions.is_date_yyyymmdd(str(to_date), silent=True):
# datetime_end=datetime.datetime.strptime(str(to_date),'%Y%m%d')
# # If it is a negative number, subtract from current date
# elif isinstance(to_date,int) or isinstance(to_date,long):
# if to_date < 0:
# datetime_end=datetime.datetime.today() - datetime.timedelta(days=-to_date)
# else:
# datetime_end=datetime.datetime.today()
# except:
# pass
#
# try:
# list_filenames = sentinelsat_api.sentinelsat_getlists(base_url, template, datetime_start, datetime_end)#frequency.get_dates(datetime_start, datetime_end)
# except Exception as inst:
# logger.debug("Error in sentinelsat.get_lists: %s" %inst.args[0])
# raise
# try:
# dates = frequency.get_dates(datetime_start, datetime_end)
# except Exception as inst:
# logger.debug("Error in frequency.get_dates: %s" %inst.args[0])
# raise
#
# try:
# list_filenames = frequency.get_internet_dates(dates, template)
# except Exception as inst:
# logger.debug("Error in frequency.get_internet_dates: %s" %inst.args[0])
# raise
return list_filenames
def test_is_date_time(self):
self.assertTrue(f.is_date_yyyymmdd(self.string_yyyymmdd))
self.assertTrue(f.is_date_mmdd(self.string_mmdd))
self.assertTrue(f.is_date_yyyymmddhhmm(self.string_yyyymmddhhmm))
self.assertTrue(f.is_date_yyyydoy(self.string_yyyydoy))
def test_is_date_time(self):
self.assertTrue(f.is_date_yyyymmdd(self.string_yyyymmdd))
self.assertTrue(f.is_date_mmdd(self.string_mmdd))
self.assertTrue(f.is_date_yyyymmddhhmm(self.string_yyyymmddhhmm))
self.assertTrue(f.is_date_yyyydoy(self.string_yyyydoy))
def build_list_matching_files_cds(base_url, template, from_date, to_date,
frequency_id, resourcename_uuid):
# Add a check on frequency
try:
frequency = datasets.Dataset.get_frequency(
frequency_id, datasets.Frequency.DATEFORMAT.DATETIME)
except Exception as inst:
logger.debug("Error in datasets.Dataset.get_frequency: %s" %
inst.args[0])
raise
# Manage the start_date (mandatory).
try:
# If it is a date, convert to datetime
if functions.is_date_yyyymmdd(str(from_date), silent=True):
datetime_start = datetime.datetime.strptime(
str(from_date), '%Y%m%d')
else:
# If it is a negative number, subtract from current date
if isinstance(from_date, int) or isinstance(from_date, int):
if from_date < 0:
datetime_start = datetime.datetime.today(
) - datetime.timedelta(days=-from_date)
else:
logger.debug("Error in Start Date: must be YYYYMMDD or -Ndays")
raise Exception("Start Date not valid")
except:
raise Exception("Start Date not valid")
# Manage the end_date (mandatory).
try:
if functions.is_date_yyyymmdd(str(to_date), silent=True):
datetime_end = datetime.datetime.strptime(str(to_date), '%Y%m%d')
# If it is a negative number, subtract from current date
elif isinstance(to_date, int) or isinstance(to_date, int):
if to_date < 0:
datetime_end = datetime.datetime.today() - datetime.timedelta(
days=-to_date)
else:
datetime_end = datetime.datetime.today()
except:
pass
try:
dates = frequency.get_dates(datetime_start, datetime_end)
except Exception as inst:
logger.debug("Error in frequency.get_dates: %s" % inst.args[0])
raise
try:
if sys.platform == 'win32':
template.replace("-", "#")
# return lst
list_input_files = cds_api.create_list_cds(dates, template, base_url,
resourcename_uuid)
except Exception as inst:
logger.debug("Error in frequency.get_internet_dates: %s" %
inst.args[0])
raise
return list_input_files
def test_is_date_yyyymmdd(self):
self.assertTrue(functions.is_date_yyyymmdd(self.string_yyyymmdd))
