def read_bufr_file(input_bufr_file):
# #[ read a bufr file
"""
read the file using the BUFRReader class and get the data
with the get_values_as_2d_array method
"""
# get an instance of the BUFR class
# which automatically opens the file for reading and decodes it
bufr = BUFRReader(input_bufr_file, warn_about_bufr_size=False)
msg_nr = -1
for msg_nr, msg in enumerate(bufr):
num_subsets = msg.get_num_subsets()
for subs, msg_or_subset_data in enumerate(msg):
#names = msg_or_subset_data.names
#units = msg_or_subset_data.units
data = msg_or_subset_data.data
if data.shape[0] == 0:
print('NO DATA FOUND! this seems an empty BUFR message !')
continue
print('loaded BUFR msg nr. ', msg_nr, 'shape = ', data.shape)
if len(data.shape) == 1:
print('data[:2] = ', data[:2])
else:
print('data[:2,:2] = ', data[:2, :2])
if subs > 1:
break
# close the file
bufr.close()
if msg_nr == -1:
print('no BUFR messages found, are you sure this is a BUFR file?')
def print_bufr_content1(input_bufr_file, output_fd, separator, max_msg_nr):
# #[ implementation 1
"""
example implementation using the BUFRReader class
combined with the get_values_as_2d_array method
"""
# get an instance of the BUFR class
# which automatically opens the file for reading and decodes it
bob = BUFRReader(input_bufr_file, warn_about_bufr_size=False)
msg_nr = 0
while True:
try:
bob.get_next_msg()
msg_nr += 1
except EOFError:
break
# add header strings
# print 'DEBUG: bob.msg_loaded ',bob.msg_loaded
if bob.msg_loaded == 1:
list_of_names = []
list_of_units = []
list_of_names.extend(bob.get_names())
list_of_units.extend(bob.get_units())
# print 'DEBUG: ',separator.join(list_of_names)
# print 'DEBUG: ',separator.join(list_of_units)
output_fd.write(separator.join(list_of_names) + "\n")
output_fd.write(separator.join(list_of_units) + "\n")
data = bob.get_values_as_2d_array()
# print 'DEBUG: data.shape = ', data.shape
if data.shape[0] * data.shape[1] == 0:
print 'NO DATA FOUND! this seems an empty BUFR message !'
continue
for subs in range(len(data[:, 0])):
output_fd.write(
str(subs) + separator +
separator.join(str(val) for val in data[subs, :]) + "\n")
print 'converted BUFR msg nr. ', msg_nr
if ((max_msg_nr > 0) and (msg_nr >= max_msg_nr)):
print 'skipping remainder of this BUFR file'
break
# close the file
bob.close()
if msg_nr == 0:
print 'no BUFR messages found, are you sure this is a BUFR file?'
def sort_msgs(input_bufr_file):
# #[
"""
a little example routine to demonstrate how to extract
BUFR messages from a BUFR file, sort them, and write them
out again to another file.
"""
# get an instance of the BUFR class
# which automatically opens the file for reading and decodes it
bob = BUFRReader(input_bufr_file, warn_about_bufr_size=False)
files_dict = {}
msg_nr = 0
while True:
try:
bob.get_next_msg()
msg_nr += 1
except EOFError:
break
print 'handling message nr ', msg_nr
list_of_unexp_descr = bob.bufr_obj.py_unexp_descr_list
output_filename = '_'.join(d for d in list_of_unexp_descr)
if files_dict.has_key(output_filename):
fdescr = files_dict[output_filename][0]
files_dict[output_filename][1] += 1 # increment count
else:
fdescr = open(output_filename, 'wb')
count = 1
files_dict[output_filename] = [fdescr, count]
fdescr.write(bob.bufr_obj.encoded_message)
generated_files = files_dict.keys()
for k in files_dict.keys():
count = files_dict[k][1]
print 'file ', k, ' contains ', count, ' messages'
files_dict[k][0].close()
return generated_files
def read_bufr_file(input_bufr_file):
# #[ read a bufr file
"""
read the file using the BUFRReader class and get the data
with the get_values_as_2d_array method
"""
# get an instance of the BUFR class
# which automatically opens the file for reading and decodes it
bufr = BUFRReader(input_bufr_file, warn_about_bufr_size=False)
msg_nr = -1
for msg_nr, msg in enumerate(bufr):
num_subsets = msg.get_num_subsets()
for subs, msg_or_subset_data in enumerate(msg):
#names = msg_or_subset_data.names
#units = msg_or_subset_data.units
data = msg_or_subset_data.data
if data.shape[0] == 0:
print('NO DATA FOUND! this seems an empty BUFR message !')
continue
print('loaded BUFR msg nr. ', msg_nr,
'shape = ', data.shape)
if len(data.shape) == 1:
print('data[:2] = ', data[:2].tolist())
else:
print('data[:2,:2] = ', data[:2,:2].tolist())
if subs > 1:
break
# close the file
bufr.close()
if msg_nr == -1:
print('no BUFR messages found, are you sure this is a BUFR file?')
def read_bufr_file(input_bufr_file):
# #[ read a bufr file
"""
read the file using the BUFRReader class and get the data
with the get_values_as_2d_array method
"""
# get an instance of the BUFR class
# which automatically opens the file for reading and decodes it
bob = BUFRReader(input_bufr_file, warn_about_bufr_size=False)
msg_nr = 0
while True:
try:
bob.get_next_msg()
msg_nr += 1
except EOFError:
break
nsubsets = bob.get_num_subsets()
data1 = bob.get_subset_values(1)
if numpy.shape(data1)[0] == 0:
print('NO DATA FOUND! this seems an empty BUFR message !')
continue
print('loaded BUFR msg nr. ', msg_nr, 'shape = ', numpy.shape(data1))
print('data1[:2] = ', data1[:2])
if nsubsets > 1:
data2 = bob.get_subset_values(2)
if numpy.shape(data2)[0] == 0:
print('NO DATA FOUND! this seems an empty BUFR message !')
continue
print('data2[:2] = ', data2[:2])
# close the file
bob.close()
if msg_nr == 0:
print('no BUFR messages found, are you sure this is a BUFR file?')
def decoding_example(input_bufr_file):
"""
wrap the example in a function to circumvent the pylint
convention of requiring capitals for constants in the global
scope (since most of these variables are not constants at all))
"""
# suppres the default ECMWF welcome message which
# is not yet redirected to the above defined fileunit
os.environ['PRINT_TABLE_NAMES'] = 'FALSE'
# read the binary data using the BUFRReader class
print('loading testfile: ', input_bufr_file)
with BUFRReader(input_bufr_file) as bufr:
for msg in bufr:
for msg_or_subset_data in msg:
data = msg_or_subset_data.data
names = msg_or_subset_data.names
units = msg_or_subset_data.units
print(data.shape)
print(len(names))
print(len(units))
# debug
print('values[:25]: ', msg.values[:25].tolist())
print('values[-25:]: ', msg.values[-25:].tolist())
msg.write_msg_to_file()
bwr.close()
##############################
# reopen the BUFR file as test
##############################
print('*'*50)
input_bufr_file = output_bufr_file
bufr = BUFRReader(input_bufr_file, warn_about_bufr_size=False)
# just 1 msg in this test file, so no looping needed
for msg in bufr:
print('num_subsets = ', msg.get_num_subsets())
for subs, msg_or_subset_data in enumerate(msg):
list_of_names = msg_or_subset_data.names
list_of_units = msg_or_subset_data.units
data = msg_or_subset_data.data
print('"subset nr"'+','+','.join(list_of_names))
print('""'+','+','.join(list_of_units))
print('data.shape = ', data.shape)
for irow, row in enumerate(data):
print('{}, {} : {}'.
format(subs+1, irow,
', '.join('{}'.format(val) for val in row.tolist())))
def print_bufr_content4(input_bufr_file, output_fd, separator, max_msg_nr,
expand_flags, expand_strings, descr_multiplier):
# #[ implementation 4
"""
example implementation using the BUFRReader class
to decode a bufr file using delayed replication.
Since these files may have different descriptor lists
for each subset, a different call pattern is needed.
"""
# get an instance of the BUFR class
# which automatically opens the file for reading and decodes it
bufr = BUFRReader(
input_bufr_file,
warn_about_bufr_size=False,
#verbose=True, expand_flags=expand_flags,
verbose=False,
expand_flags=expand_flags,
expand_strings=expand_strings,
descr_multiplyer=descr_multiplier)
msg_nr = -1
for msg_nr, msg in enumerate(bufr):
# since this example assumes a bufr file using delayed replication
# always request and add the header for each subset
nsubsets = msg.get_num_subsets()
# print('nsubsets = ', nsubsets)
for msg_or_subs_nr, msg_or_subset_data in enumerate(msg):
# add header strings
list_of_names = msg_or_subset_data.names
list_of_units = msg_or_subset_data.units
data = msg_or_subset_data.data
if numpy.shape(data)[0] == 0:
print('NO DATA FOUND! this seems an empty BUFR message !')
continue
output_fd.write('"subset nr"' + separator +
separator.join(list_of_names) + "\n")
output_fd.write('""' + separator + separator.join(list_of_units) +
"\n")
# print(data.shape)
if len(data.shape) == 1:
# we are walking over subsets
subs = msg_or_subs_nr
output_fd.write(
str(subs + 1) + separator +
separator.join(to_str(val) for val in data[:]) + "\n")
else:
# we are getting a 2D array as result
for subs in range(data.shape[0]):
output_fd.write(
str(subs + 1) + separator +
separator.join(to_str(val)
for val in data[subs, :]) + "\n")
print('converted BUFR msg nr. ', msg_nr + 1)
if (max_msg_nr >= 0) and (msg_nr >= max_msg_nr):
print('skipping remainder of this BUFR file')
break
# close the file
bufr.close()
if msg_nr == -1:
print('no BUFR messages found, are you sure this is a BUFR file?')
def select_subsets(input_bufr_file, output_bufr_file):
# #[ select on subsets
"""
select data and write out again
"""
# get an instance of the BUFR class
# which automatically opens the file for reading and decodes it
bob = BUFRReader(input_bufr_file, warn_about_bufr_size=False)
# open the file for writing
rbf_out = RawBUFRFile()
rbf_out.open(output_bufr_file, 'wb')
msg_nr = 0
while True:
try:
bob.get_next_msg()
msg_nr += 1
except EOFError:
break
data = bob.get_values_as_2d_array()
print('data.shape = ', data.shape)
if data.shape[0] * data.shape[1] == 0:
print('NO DATA FOUND! this seems an empty BUFR message !')
continue
# select every other subset
new_data = data[::2, :]
print('new_data.shape = ', new_data.shape)
#bob.bufr_obj
nsub = bob.bufr_obj.get_num_subsets()
n_exp_descr = len(bob.bufr_obj.values) / nsub
bob.bufr_obj.fill_descriptor_list(
nr_of_expanded_descriptors=n_exp_descr)
bob.bufr_obj.ktdlst = bob.bufr_obj.get_descriptor_list()
delayed_repl_data = bob.bufr_obj.derive_delayed_repl_factors()
bob.bufr_obj.fill_delayed_repl_data(delayed_repl_data)
new_nsub = new_data.shape[0]
bob.bufr_obj.nr_subsets = new_nsub
btm = BufrTemplate()
btm.add_descriptors(*bob.bufr_obj.ktdlst) #[:self.ktdlen])
btm.nr_of_delayed_repl_factors = 1
btm.del_repl_max_nr_of_repeats_list = list(delayed_repl_data)
bob.bufr_obj.register_and_expand_descriptors(btm)
# activate this one if the encoding crashes without clear cause:
# bob.bufr_obj.estimated_num_bytes_for_encoding = 25000
bob.bufr_obj.kdate = new_nsub * list(delayed_repl_data)
print('bob.bufr_obj.cvals.shape = ', bob.bufr_obj.cvals.shape)
bob.bufr_obj.encode_data(new_data, bob.bufr_obj.cvals[:32, :])
rbf_out.write_raw_bufr_msg(bob.bufr_obj.encoded_message)
#for subs in range(len(data[:, 0])):
# output_fd.write(str(subs)+separator+
# separator.join(str(val) for val in data[subs, :])+
# "\n")
print('converted BUFR msg nr. ', msg_nr)
# close the file
bob.close()
if msg_nr == 0:
print('no BUFR messages found, are you sure this is a BUFR file?')
rbf_out.close()
# debug
print('values[:25]: ', msg.values[:25].tolist())
print('values[-25:]: ', msg.values[-25:].tolist())
msg.write_msg_to_file()
bwr.close()
##############################
# reopen the BUFR file as test
##############################
print('*' * 50)
input_bufr_file = output_bufr_file
bufr = BUFRReader(input_bufr_file,
expand_strings=True,
warn_about_bufr_size=False)
# just 1 msg in this test file, so no looping needed
for msg in bufr:
print('num_subsets = ', msg.get_num_subsets())
for subs, msg_or_subset_data in enumerate(msg):
list_of_names = msg_or_subset_data.names
list_of_units = msg_or_subset_data.units
data = msg_or_subset_data.data
print('"subset nr"' + ',' + ','.join(list_of_names))
print('""' + ',' + ','.join(list_of_units))
print(str(subs + 1) + ',' + ','.join(str(val) for val in data[:]))
bufr.close()
def print_bufr_content1(input_bufr_file, output_fd, separator,
max_msg_nr, expand_flags):
# #[ implementation 1
"""
example implementation using the BUFRReader class
combined with the get_values_as_2d_array method
"""
# get an instance of the BUFR class
# which automatically opens the file for reading and decodes it
bob = BUFRReader(input_bufr_file, warn_about_bufr_size=False,
expand_flags=expand_flags)
msg_nr = 0
while True:
try:
bob.get_next_msg()
msg_nr += 1
except EOFError:
break
# add header strings
# print('DEBUG: bob.msg_loaded ',bob.msg_loaded)
list_of_names = []
list_of_units = []
list_of_names.extend(bob.get_names())
list_of_units.extend(bob.get_units())
list_of_unexp_descr = bob.get_unexp_descr_list()
#print('list_of_names = ',list_of_names)
#print('list_of_units = ',list_of_units)
if bob.msg_loaded == 1:
output_fd.write('"subset nr"'+separator)
if list_of_names:
for name in list_of_names[:-1]:
output_fd.write('"'+name+'"'+separator)
name = list_of_names[-1]
output_fd.write('"'+name+'"\n')
else:
output_fd.write('"[NO DATA]"\n')
output_fd.write('""'+separator)
if list_of_units:
for unit in list_of_units[:-1]:
output_fd.write('"'+unit+'"'+separator)
unit = list_of_units[-1]
output_fd.write('"'+unit+'"\n')
else:
output_fd.write('"[NO DATA]"\n')
list_of_unexp_descr_first_msg = bob.bufr_obj.py_unexp_descr_list
#print('list_of_unexp_descr_first_msg = ',
# list_of_unexp_descr_first_msg)
data = bob.get_values_as_2d_array()
if list_of_unexp_descr != list_of_unexp_descr_first_msg:
print('\n\n')
print('WARNING: it seems different types of BUFR messages')
print('are mixed in this BUFR file, meaning that the list of')
print('descriptor names and units printed on the first 2 output')
print('lines will not match with all lines of data.')
print('To prevent confusion, therefore decoding is halted')
print('It is recommended to first sort BUFR messages by type')
print('before converting them to ascii or csv.')
print('The example script sort_bufr_msgs.py can be used')
print('to sort a BUFR file.')
print('\n\n')
print('Detailed info:')
print('list_of_unexp_descr != list_of_unexp_descr_first_msg !')
print('list_of_unexp_descr = ',
list_of_unexp_descr)
print('list_of_unexp_descr_first_msg = ',
list_of_unexp_descr_first_msg)
sys.exit(0)
if data.shape[0]*data.shape[1] == 0:
print('NO DATA FOUND! this seems an empty BUFR message !')
continue
for subs in range(len(data[:, 0])):
output_fd.write(str(subs+1)+separator+
separator.join(str(val) for val in data[subs, :])+
"\n")
print('converted BUFR msg nr. ', msg_nr)
if (max_msg_nr > 0) and (msg_nr >= max_msg_nr):
print('skipping remainder of this BUFR file')
break
# close the file
bob.close()
if msg_nr == 0:
print('no BUFR messages found, are you sure this is a BUFR file?')
def print_bufr_content5(input_bufr_file, output_fd, separator,
max_msg_nr, expand_flags):
# #[ implementation 5
"""
example implementation using the BUFRReader class
to decode a bufr file using delayed replication.
Since these files may have different descriptor lists
for each subset, a different call pattern is needed.
"""
# testcases:
# ./example_programs/bufr_to_ascii.py -5 -c -o tmp.csv \
# -i ./pybufr_ecmwf/ecmwf_bufr_lib/bufrdc_000403/data/syno_1.bufr
#
# ./example_programs/bufr_to_ascii.py -5 -c -o tmp.csv \
# -i ../BUFR_test_files/synop_knmi_via_ko_janssen/MSSAEOL_00002950.b
#
names_to_be_selected = ['temperature', 'wind']
names_to_be_excluded = ['minimum', 'maximum']
write_names_and_units_just_once = True
# get an instance of the BUFR class
# which automatically opens the file for reading and decodes it
bob = BUFRReader(input_bufr_file, warn_about_bufr_size=False,
verbose=False, expand_flags=expand_flags)
msg_nr = 0
not_yet_printed = True
while True:
try:
bob.get_next_msg()
msg_nr += 1
except EOFError:
break
# since this example assumes a bufr file using delayed replication
# always request and add the header for each subset
nsubsets = bob.get_num_subsets()
for subs in range(1, nsubsets+1):
print('==> subset ', subs)
# add header strings
(list_of_names, list_of_units) = bob.get_names_and_units(subs)
data = bob.get_subset_values(subs) #,autoget_cval=True)
selected_names = []
selected_units = []
selected_values = []
for i, name in enumerate(list_of_names):
selected = False
for name in names_to_be_selected:
if name in name.lower():
selected = True
for name in names_to_be_excluded:
if name in name.lower():
selected = False
if selected:
# print(' '*10,name,'=',data[i],list_of_units[i])
selected_names.append(list_of_names[i])
selected_units.append(list_of_units[i])
selected_values.append(data[i])
if len(selected_values) == 0:
print('NO DATA SELECTED for BUFR message %d and subset %d!' %
(msg_nr, subs))
continue
if write_names_and_units_just_once and not_yet_printed:
output_fd.write('"subset nr"'+separator+
separator.join(selected_names) + "\n")
output_fd.write('""'+separator+
separator.join(selected_units) + "\n")
not_yet_printed = False
output_fd.write(str(subs)+separator+
separator.join(str(val) for val in selected_values)+
"\n")
print('='*25)
print('converted BUFR msg nr. ', msg_nr)
print('='*25)
if (max_msg_nr > 0) and (msg_nr >= max_msg_nr):
print('skipping remainder of this BUFR file')
break
# close the file
bob.close()
if msg_nr == 0:
print('no BUFR messages found, are you sure this is a BUFR file?')
def sort_msgs(input_bufr_file):
# #[
"""
a little example routine to demonstrate how to extract
BUFR messages from a BUFR file, sort them, and write them
out again to another file.
"""
# get an instance of the BUFR class
# which automatically opens the file for reading and decodes it
bob = BUFRReader(input_bufr_file, warn_about_bufr_size=False)
num_msgs = bob.rbf.get_num_bufr_msgs()
progress_step = max(1, int(num_msgs / 20))
files_dict = {}
msg_nr = 0
while True:
can_be_decoded = False
try:
bob.get_next_msg()
can_be_decoded = True
except EOFError:
break
except KeyError:
# allow sorting of BUFR messages that cannot be decoded
# because the needed key is not in the available set of
# BUFR table files. This tool only uses unexpanded descriptors
# so ability to decode should not be required.
# A user still may wish to first sort, and then decode
# a subset of messages that can be decoded.
pass
msg_nr += 1
if progress_step * int(msg_nr / progress_step) == msg_nr:
print('handling message nr {} out of {}'.format(msg_nr, num_msgs))
list_of_unexp_descr = bob.bufr_obj.py_unexp_descr_list
output_filename = construct_unique_filename(list_of_unexp_descr)
if files_dict.has_key(output_filename):
fdescr = files_dict[output_filename][0]
files_dict[output_filename][1] += 1 # increment count
else:
fdescr = open(output_filename, 'wb')
count = 1
files_dict[output_filename] = [
fdescr, count, can_be_decoded, list_of_unexp_descr
]
fdescr.write(bob.bufr_obj.encoded_message)
generated_files = files_dict.keys()
num_that_can_be_decoded = 0
num_that_cannot_be_decoded = 0
for k in files_dict:
fdescr, count, can_be_decoded, list_of_unexp_descr = files_dict[k]
print('file {} contains {} messages'.format(k[:25], count))
files_dict[k][0].close()
if can_be_decoded:
num_that_can_be_decoded += 1
else:
num_that_cannot_be_decoded += 1
# check to see if local descriptors are present
for descr in list_of_unexp_descr:
if int(descr[3:]) >= 192:
print('==>A local descriptor definition is present: ',
descr)
print('==>this template cannot be decoded with ' +
'standard WMO BUFR tables.')
print('Sorting results:')
print('BUFR messages with {} different templates are present in this file'.
format(num_that_can_be_decoded + num_that_cannot_be_decoded))
if num_that_cannot_be_decoded > 0:
print('decoding is not possible for {} templates.'.format(
num_that_cannot_be_decoded))
return generated_files
def print_bufr_content1(input_bufr_file, output_fd, separator,
max_msg_nr, expand_flags):
# #[ implementation 1
"""
example implementation using the BUFRReader class
combined with the get_values_as_2d_array method
"""
# get an instance of the BUFR class
# which automatically opens the file for reading and decodes it
bufr = BUFRReader(input_bufr_file, warn_about_bufr_size=False,
expand_flags=expand_flags)
msg_nr = -1
for msg_nr, msg in enumerate(bufr):
num_subsets = msg.get_num_subsets()
list_of_unexp_descr = msg.get_unexp_descr_list()
for subs, msg_or_subset_data in enumerate(msg):
# get the actual values
data = msg_or_subset_data.data
if msg_nr == 0 and subs == 0:
list_of_unexp_descr_first_msg = list_of_unexp_descr[:]
# add header strings
list_of_names = msg_or_subset_data.names
list_of_units = msg_or_subset_data.units
output_fd.write('"subset nr"'+separator)
if list_of_names:
for name in list_of_names[:-1]:
output_fd.write('"'+name+'"'+separator)
name = list_of_names[-1]
output_fd.write('"'+name+'"\n')
else:
output_fd.write('"[NO DATA]"\n')
output_fd.write('""'+separator)
if list_of_units:
for unit in list_of_units[:-1]:
output_fd.write('"'+unit+'"'+separator)
unit = list_of_units[-1]
output_fd.write('"'+unit+'"\n')
else:
output_fd.write('"[NO DATA]"\n')
try:
data_is_2d = True
if len(numpy.shape(data)) == 1:
data_is_2d = False
if data_is_2d:
ns = numpy.shape(data)[0]
for subs_cnt in range(ns):
output_fd.write(str(subs_cnt+1)+separator+
separator.join(str(val)
for val in
data[subs_cnt, :])+
"\n")
else:
# 1D data is returned if character values may be present
# in the data (in case autoget_cval or expand_flags
# are active)
# it may also happen if the message uses delayed
# replication and has variable lengths for the
# different subsets
subs_cnt = msg_or_subset_data.current_subset
output_fd.write(str(subs_cnt)+separator+
separator.join(str(val)
for val in
data[:])+
"\n")
except TypeError:
# in case of delayed replication or when character strings are
# present (i.e. expand_flags = True) data will be returned as
# a 1D list in stead of a 2D numpy array.
# This generates a TypeError in the data[subs, :] indexing above
output_fd.write(str(subs+1)+separator+
separator.join(str(val) for val in data[:])+
"\n")
if list(list_of_unexp_descr) != list(list_of_unexp_descr_first_msg):
print('\n\n')
print('ERROR: it seems different types of BUFR messages')
print('are mixed in this BUFR file, meaning that the list of')
print('descriptor names and units printed on the first 2 output')
print('lines will not match with all lines of data.')
print('To prevent confusion, therefore decoding is halted')
print('It is recommended to first sort BUFR messages by type')
print('before converting them to ascii or csv.')
print('The example script sort_bufr_msgs.py can be used')
print('to sort a BUFR file.')
print('\n\n')
print('Detailed info:')
print('list_of_unexp_descr != list_of_unexp_descr_first_msg !')
print('list_of_unexp_descr = ',
list_of_unexp_descr)
print('list_of_unexp_descr_first_msg = ',
list_of_unexp_descr_first_msg)
sys.exit(1)
if numpy.shape(data)[0] == 0:
print('NO DATA FOUND! this seems an empty BUFR message !')
continue
print('converted BUFR msg nr. ', msg_nr+1)
if msg_nr >= max_msg_nr:
print('skipping remainder of this BUFR file')
break
# close the file
bufr.close()
if msg_nr == -1:
print('no BUFR messages found, are you sure this is a BUFR file?')
def print_bufr_content5(input_bufr_file, output_fd, separator, max_msg_nr,
expand_flags):
# #[ implementation 5
"""
example implementation using the BUFRReader class
to decode a bufr file using delayed replication.
Since these files may have different descriptor lists
for each subset, a different call pattern is needed.
"""
# testcases:
# ./example_programs/bufr_to_ascii.py -5 -c -o tmp.csv \
# -i ./pybufr_ecmwf/ecmwf_bufr_lib/bufrdc_000403/data/syno_1.bufr
#
# ./example_programs/bufr_to_ascii.py -5 -c -o tmp.csv \
# -i ../BUFR_test_files/synop_knmi_via_ko_janssen/MSSAEOL_00002950.b
#
names_to_be_selected = ['temperature', 'wind']
names_to_be_excluded = ['minimum', 'maximum']
write_names_and_units_just_once = True
# get an instance of the BUFR class
# which automatically opens the file for reading and decodes it
bob = BUFRReader(input_bufr_file,
warn_about_bufr_size=False,
verbose=False,
expand_flags=expand_flags)
msg_nr = 0
not_yet_printed = True
while True:
try:
bob.get_next_msg()
msg_nr += 1
except EOFError:
break
# since this example assumes a bufr file using delayed replication
# always request and add the header for each subset
nsubsets = bob.get_num_subsets()
for subs in range(1, nsubsets + 1):
print('==> subset ', subs)
# add header strings
(list_of_names, list_of_units) = bob.get_names_and_units(subs)
data = bob.get_subset_values(subs) #,autoget_cval=True)
selected_names = []
selected_units = []
selected_values = []
for i, name in enumerate(list_of_names):
selected = False
for name in names_to_be_selected:
if name in name.lower():
selected = True
for name in names_to_be_excluded:
if name in name.lower():
selected = False
if selected:
# print(' '*10,name,'=',data[i],list_of_units[i])
selected_names.append(list_of_names[i])
selected_units.append(list_of_units[i])
selected_values.append(data[i])
if len(selected_values) == 0:
print('NO DATA SELECTED for BUFR message %d and subset %d!' %
(msg_nr, subs))
continue
if write_names_and_units_just_once and not_yet_printed:
output_fd.write('"subset nr"' + separator +
separator.join(selected_names) + "\n")
output_fd.write('""' + separator +
separator.join(selected_units) + "\n")
not_yet_printed = False
output_fd.write(
str(subs) + separator +
separator.join(str(val) for val in selected_values) + "\n")
print('=' * 25)
print('converted BUFR msg nr. ', msg_nr)
print('=' * 25)
if (max_msg_nr > 0) and (msg_nr >= max_msg_nr):
print('skipping remainder of this BUFR file')
break
# close the file
bob.close()
if msg_nr == 0:
print('no BUFR messages found, are you sure this is a BUFR file?')
if ((max_msg_nr > 0) and (msg_nr >= max_msg_nr)):
print 'skipping remainder of this BUFR file'
break
# close the file
bob.close()
if msg_nr == 0:
print 'no BUFR messages found, are you sure this is a BUFR file?'
# #]
if __name__ == '__main__':
bufrdir = '/net/bhw379/nobackup/users/plas/temp/BUFRdata/'
bufrfile = os.path.join(bufrdir, 'BUFR.radarv.bewid')
os.environ['BUFR_TABLES'] = os.path.join(bufrdir, 'BUFRtables')
bob = BUFRReader(bufrfile, warn_about_bufr_size=False)
bob.get_next_msg()
names = bob.get_names()
units = bob.get_units()
print set(names)
print 20 * '='
print set(units)
for n, u in zip(names, units):
if "dB" in u: print n, u
outf = open('radarBUFR.dat', 'w')
print_bufr_content1(bufrfile, outf, ',', 5)
def sort_msgs(input_bufr_file):
# #[
"""
a little example routine to demonstrate how to extract
BUFR messages from a BUFR file, sort them, and write them
out again to another file.
"""
# get an instance of the BUFR class
# which automatically opens the file for reading and decodes it
bob = BUFRReader(input_bufr_file, warn_about_bufr_size=False)
num_msgs = bob.rbf.get_num_bufr_msgs()
progress_step = max(1,int(num_msgs/20))
files_dict = {}
msg_nr = 0
while True:
can_be_decoded = False
try:
bob.get_next_msg()
can_be_decoded = True
except EOFError:
break
except KeyError:
# allow sorting of BUFR messages that cannot be decoded
# because the needed key is not in the available set of
# BUFR table files. This tool only uses unexpanded descriptors
# so ability to decode should not be required.
# A user still may wish to first sort, and then decode
# a subset of messages that can be decoded.
pass
msg_nr += 1
if progress_step*int(msg_nr/progress_step) == msg_nr:
print('handling message nr {} out of {}'.format(msg_nr, num_msgs))
list_of_unexp_descr = bob.bufr_obj.py_unexp_descr_list
output_filename = construct_unique_filename(list_of_unexp_descr)
if files_dict.has_key(output_filename):
fdescr = files_dict[output_filename][0]
files_dict[output_filename][1] += 1 # increment count
else:
fdescr = open(output_filename, 'wb')
count = 1
files_dict[output_filename] = [fdescr, count, can_be_decoded,
list_of_unexp_descr]
fdescr.write(bob.bufr_obj.encoded_message)
generated_files = files_dict.keys()
num_that_can_be_decoded = 0
num_that_cannot_be_decoded = 0
for k in files_dict:
fdescr, count, can_be_decoded, list_of_unexp_descr = files_dict[k]
print('file {} contains {} messages'.format(k[:25], count))
files_dict[k][0].close()
if can_be_decoded:
num_that_can_be_decoded += 1
else:
num_that_cannot_be_decoded += 1
# check to see if local descriptors are present
for d in list_of_unexp_descr:
if int(d[3:]) >= 192:
print('==>A local descriptor definition is present: ', d)
print('==>this template cannot be decoded with '+
'standard WMO BUFR tables.')
print('Sorting results:')
print('BUFR messages with {} different templates are present in this file'.
format(num_that_can_be_decoded+num_that_cannot_be_decoded))
if num_that_cannot_be_decoded > 0:
print('decoding is not possible for {} templates.'.
format(num_that_cannot_be_decoded))
return generated_files
def reopen_bufr_file(input_bufr_file):
# #[ open bufr file
'''
open a bufr file for reading and print its content
'''
print '*'*50
from pybufr_ecmwf.bufr import BUFRReader
bob = BUFRReader(input_bufr_file, warn_about_bufr_size=False)
bob.setup_tables(table_b_to_use='B'+TABLE_NAME,
table_d_to_use='D'+TABLE_NAME)
bob.get_next_msg()
print 'num_subsets: ', bob.get_num_subsets()
if USE_DELAYED_REPLICATION:
data1 = bob.get_subset_values(0)
print 'data1 = ', data1
data2 = bob.get_subset_values(1)
print 'data2 = ', data2
else:
print 'num_elements: ', bob.get_num_elements()
print bob.get_names()
print bob.get_units()
data = bob.get_values_as_2d_array()
print data.shape
print data
print 'bob.bufr_obj.values = '
print bob.bufr_obj.values
textdata = bob.get_value(3, 0)
print 'textdata(3,0)', textdata
textdata = bob.get_value(3, 0, get_cval=True)
print 'textdata(3,0)', textdata
textdata = bob.get_values(3, get_cval=True)
print 'textdata(3,:)', textdata
bob.close()
def select_subsets(input_bufr_file, output_bufr_file):
# #[ select on subsets
"""
select data and write out again
"""
# get an instance of the BUFR class
# which automatically opens the file for reading and decodes it
bob = BUFRReader(input_bufr_file, warn_about_bufr_size=False)
# open the file for writing
rbf_out = RawBUFRFile()
rbf_out.open(output_bufr_file, 'wb')
msg_nr = 0
while True:
try:
bob.get_next_msg()
msg_nr += 1
except EOFError:
break
data = bob.get_values_as_2d_array()
print 'data.shape = ', data.shape
if data.shape[0]*data.shape[1] == 0:
print 'NO DATA FOUND! this seems an empty BUFR message !'
continue
# select every other subset
new_data = data[::2, :]
print 'new_data.shape = ', new_data.shape
#bob.bufr_obj
nsub = bob.bufr_obj.get_num_subsets()
n_exp_descr = len(bob.bufr_obj.values)/nsub
bob.bufr_obj.fill_descriptor_list(nr_of_expanded_descriptors=
n_exp_descr)
bob.bufr_obj.ktdlst = bob.bufr_obj.get_descriptor_list()
delayed_repl_data = bob.bufr_obj.derive_delayed_repl_factors()
bob.bufr_obj.fill_delayed_repl_data(delayed_repl_data)
new_nsub = new_data.shape[0]
bob.bufr_obj.nr_subsets = new_nsub
btm = BufrTemplate()
btm.add_descriptors(*bob.bufr_obj.ktdlst)#[:self.ktdlen])
btm.nr_of_delayed_repl_factors = 1
btm.del_repl_max_nr_of_repeats_list = list(delayed_repl_data)
bob.bufr_obj.register_and_expand_descriptors(btm)
# activate this one if the encoding crashes without clear cause:
# bob.bufr_obj.estimated_num_bytes_for_encoding = 25000
bob.bufr_obj.kdate = new_nsub*list(delayed_repl_data)
print 'bob.bufr_obj.cvals.shape = ', bob.bufr_obj.cvals.shape
bob.bufr_obj.encode_data(new_data, bob.bufr_obj.cvals[:32, :])
rbf_out.write_raw_bufr_msg(bob.bufr_obj.encoded_message)
#for subs in range(len(data[:, 0])):
# output_fd.write(str(subs)+separator+
# separator.join(str(val) for val in data[subs, :])+
# "\n")
print 'converted BUFR msg nr. ', msg_nr
# close the file
bob.close()
if msg_nr == 0:
print 'no BUFR messages found, are you sure this is a BUFR file?'
rbf_out.close()
def print_bufr_content1(input_bufr_file, output_fd, separator, max_msg_nr,
expand_flags):
# #[ implementation 1
"""
example implementation using the BUFRReader class
combined with the get_values_as_2d_array method
"""
# get an instance of the BUFR class
# which automatically opens the file for reading and decodes it
bufr = BUFRReader(input_bufr_file,
warn_about_bufr_size=False,
expand_flags=expand_flags)
msg_nr = -1
for msg_nr, msg in enumerate(bufr):
num_subsets = msg.get_num_subsets()
list_of_unexp_descr = msg.get_unexp_descr_list()
for subs, msg_or_subset_data in enumerate(msg):
# get the actual values
data = msg_or_subset_data.data
if msg_nr == 0 and subs == 0:
list_of_unexp_descr_first_msg = list_of_unexp_descr[:]
# add header strings
list_of_names = msg_or_subset_data.names
list_of_units = msg_or_subset_data.units
output_fd.write('"subset nr"' + separator)
if list_of_names:
for name in list_of_names[:-1]:
output_fd.write('"' + name + '"' + separator)
name = list_of_names[-1]
output_fd.write('"' + name + '"\n')
else:
output_fd.write('"[NO DATA]"\n')
output_fd.write('""' + separator)
if list_of_units:
for unit in list_of_units[:-1]:
output_fd.write('"' + unit + '"' + separator)
unit = list_of_units[-1]
output_fd.write('"' + unit + '"\n')
else:
output_fd.write('"[NO DATA]"\n')
try:
data_is_2d = True
if len(numpy.shape(data)) == 1:
data_is_2d = False
if data_is_2d:
ns = numpy.shape(data)[0]
for subs_cnt in range(ns):
output_fd.write(
str(subs_cnt + 1) + separator + separator.join(
str(val) for val in data[subs_cnt, :]) + "\n")
else:
# 1D data is returned if character values may be present
# in the data (in case autoget_cval or expand_flags
# are active)
# it may also happen if the message uses delayed
# replication and has variable lengths for the
# different subsets
subs_cnt = msg_or_subset_data.current_subset
output_fd.write(
str(subs_cnt) + separator +
separator.join(str(val) for val in data[:]) + "\n")
except TypeError:
# in case of delayed replication or when character strings are
# present (i.e. expand_flags = True) data will be returned as
# a 1D list in stead of a 2D numpy array.
# This generates a TypeError in the data[subs, :] indexing above
output_fd.write(
str(subs + 1) + separator +
separator.join(str(val) for val in data[:]) + "\n")
if list(list_of_unexp_descr) != list(list_of_unexp_descr_first_msg):
print('\n\n')
print('ERROR: it seems different types of BUFR messages')
print('are mixed in this BUFR file, meaning that the list of')
print('descriptor names and units printed on the first 2 output')
print('lines will not match with all lines of data.')
print('To prevent confusion, therefore decoding is halted')
print('It is recommended to first sort BUFR messages by type')
print('before converting them to ascii or csv.')
print('The example script sort_bufr_msgs.py can be used')
print('to sort a BUFR file.')
print('\n\n')
print('Detailed info:')
print('list_of_unexp_descr != list_of_unexp_descr_first_msg !')
print('list_of_unexp_descr = ', list_of_unexp_descr)
print('list_of_unexp_descr_first_msg = ',
list_of_unexp_descr_first_msg)
sys.exit(1)
if numpy.shape(data)[0] == 0:
print('NO DATA FOUND! this seems an empty BUFR message !')
continue
print('converted BUFR msg nr. ', msg_nr + 1)
if (max_msg_nr > 0) and (msg_nr >= max_msg_nr):
print('skipping remainder of this BUFR file')
break
# close the file
bufr.close()
if msg_nr == -1:
print('no BUFR messages found, are you sure this is a BUFR file?')
def reopen_bufr_file(input_bufr_file):
# #[ open bufr file
'''
open a bufr file for reading and print its content
'''
print '*'*50
from pybufr_ecmwf.bufr import BUFRReader
bob = BUFRReader(input_bufr_file, warn_about_bufr_size=False)
bob.setup_tables(table_b_to_use='B'+TABLE_NAME,
table_d_to_use='D'+TABLE_NAME)
bob.get_next_msg()
print 'num_subsets: ', bob.get_num_subsets()
if USE_DELAYED_REPLICATION:
data1 = bob.get_subset_values(0)
print 'data1 = ', data1
data2 = bob.get_subset_values(1)
print 'data2 = ', data2
else:
print 'num_elements: ', bob.get_num_elements()
print bob.get_names()
print bob.get_units()
data = bob.get_values_as_2d_array()
print data.shape
print data
print 'bob.bufr_obj.values = '
print bob.bufr_obj.values
textdata = bob.get_value(3, 0)
print 'textdata(3,0)', textdata
textdata = bob.get_value(3, 0, get_cval=True)
print 'textdata(3,0)', textdata
textdata = bob.get_values(3, get_cval=True)
print 'textdata(3,:)', textdata
bob.close()
def print_bufr_content2(input_bufr_file, output_fd, separator,
max_msg_nr, expand_flags):
# #[ implementation 2
"""
example implementation using the BUFRReader class
combined with the get_value method
"""
# get an instance of the BUFR class
# which automatically opens the file for reading and decodes it
bob = BUFRReader(input_bufr_file, expand_flags=expand_flags)
msg_nr = 0
while True:
try:
bob.get_next_msg()
msg_nr += 1
except EOFError:
break
# add header strings
if bob.msg_loaded == 1:
list_of_names = bob.get_names()
list_of_units = bob.get_units()
output_fd.write('"subset nr"'+separator)
if list_of_names:
for name in list_of_names[:-1]:
output_fd.write('"'+name+'"'+separator)
name = list_of_names[-1]
output_fd.write('"'+name+'"\n')
else:
output_fd.write('"[NO DATA]"\n')
output_fd.write('""'+separator)
if list_of_units:
for unit in list_of_units[:-1]:
output_fd.write('"'+unit+'"'+separator)
unit = list_of_units[-1]
output_fd.write('"'+unit+'"\n')
else:
output_fd.write('"[NO DATA]"\n')
nsubsets = bob.get_num_subsets()
for subs in range(1, nsubsets+1):
nelements = bob.get_num_elements()
data_list = []
for descr_nr in range(nelements):
data = bob.get_value(descr_nr, subs, autoget_cval=True)
data_list.append(data)
output_fd.write(str(subs)+separator+
separator.join(str(val) for val in data_list)+
"\n")
print('converted BUFR msg nr. ', msg_nr)
if (max_msg_nr > 0) and (msg_nr >= max_msg_nr):
print('skipping remainder of this BUFR file')
break
# close the file
bob.close()
if msg_nr == 0:
print('no BUFR messages found, are you sure this is a BUFR file?')
def print_bufr_content4(input_bufr_file, output_fd, separator,
max_msg_nr, expand_flags):
# #[ implementation 4
"""
example implementation using the BUFRReader class
to decode a bufr file using delayed replication.
Since these files may have different descriptor lists
for each subset, a different call pattern is needed.
"""
# get an instance of the BUFR class
# which automatically opens the file for reading and decodes it
bob = BUFRReader(input_bufr_file, warn_about_bufr_size=False,
# verbose=True, expand_flags=expand_flags)
verbose=False, expand_flags=expand_flags)
msg_nr = 0
while True:
try:
bob.get_next_msg()
msg_nr += 1
except EOFError:
break
# since this example assumes a bufr file using delayed replication
# always request and add the header for each subset
nsubsets = bob.get_num_subsets()
for subs in range(1, nsubsets+1):
# add header strings
(list_of_names, list_of_units) = bob.get_names_and_units(subs)
# currently not used
# list_of_unexp_descr = bob.bufr_obj.py_unexp_descr_list
data = bob.get_subset_values(subs, autoget_cval=True)
# print('len(list_of_names) = ', len(list_of_names))
# print('len(list_of_units) = ', len(list_of_units))
# print('len(data) = ', len(data))
if numpy.shape(data)[0] == 0:
print('NO DATA FOUND! this seems an empty BUFR message !')
continue
output_fd.write('"subset nr"'+separator+
separator.join(list_of_names) + "\n")
output_fd.write('""'+separator+
separator.join(list_of_units) + "\n")
output_fd.write(str(subs)+separator+
separator.join(str(val) for val in data[:])+
"\n")
print('converted BUFR msg nr. ', msg_nr)
if (max_msg_nr > 0) and (msg_nr >= max_msg_nr):
print('skipping remainder of this BUFR file')
break
# close the file
bob.close()
if msg_nr == 0:
print('no BUFR messages found, are you sure this is a BUFR file?')
msg['LATI'] = [55.2, 66.3, 77.4]
msg['LONG'] = [5.1, 6.2, 7.3]
# debug
print('values: ', list(msg.values))
msg.write_msg_to_file()
bwr.close()
##############################
# reopen the BUFR file as test
##############################
print('*' * 50)
input_bufr_file = output_bufr_file
bufr = BUFRReader(input_bufr_file, warn_about_bufr_size=False)
# just 1 msg in this test file, so no looping needed
for msg in bufr:
print('num_subsets = ', msg.get_num_subsets())
for subs, msg_or_subset_data in enumerate(msg):
list_of_names = msg_or_subset_data.names
list_of_units = msg_or_subset_data.units
data = msg_or_subset_data.data
print('"subset nr"' + ',' + ','.join(list_of_names))
print('""' + ',' + ','.join(list_of_units))
print(str(subs + 1) + ',' + ','.join(str(val) for val in data[:]))
bufr.close()
def print_bufr_content4(input_bufr_file, output_fd, separator,
max_msg_nr, expand_flags, expand_strings,
descr_multiplier):
# #[ implementation 4
"""
example implementation using the BUFRReader class
to decode a bufr file using delayed replication.
Since these files may have different descriptor lists
for each subset, a different call pattern is needed.
"""
# get an instance of the BUFR class
# which automatically opens the file for reading and decodes it
bufr = BUFRReader(input_bufr_file, warn_about_bufr_size=False,
#verbose=True, expand_flags=expand_flags,
verbose=False, expand_flags=expand_flags,
expand_strings=expand_strings)
bufr.tune_decoding_parameters(nr_of_descriptors_multiplier=descr_multiplier)
msg_nr = -1
for msg_nr, msg in enumerate(bufr):
# since this example assumes a bufr file using delayed replication
# always request and add the header for each subset
nsubsets = msg.get_num_subsets()
# print('nsubsets = ', nsubsets)
for msg_or_subs_nr, msg_or_subset_data in enumerate(msg):
# add header strings
list_of_names = msg_or_subset_data.names
list_of_units = msg_or_subset_data.units
data = msg_or_subset_data.data
if numpy.shape(data)[0] == 0:
print('NO DATA FOUND! this seems an empty BUFR message !')
continue
output_fd.write('"subset nr"'+separator+
separator.join(list_of_names) + "\n")
output_fd.write('""'+separator+
separator.join(list_of_units) + "\n")
# print(data.shape)
if len(data.shape) == 1:
# we are walking over subsets
subs = msg_or_subs_nr
output_fd.write(str(subs+1)+separator+
separator.join(to_str(val)
for val in data[:])+
"\n")
else:
# we are getting a 2D array as result
for subs in range(data.shape[0]):
output_fd.write(str(subs+1)+separator+
separator.join(to_str(val)
for val in data[subs, :])+
"\n")
print('converted BUFR msg nr. ', msg_nr+1)
if msg_nr >= max_msg_nr:
print('skipping remainder of this BUFR file')
break
# close the file
bufr.close()
if msg_nr == -1:
print('no BUFR messages found, are you sure this is a BUFR file?')
