def __init__(self,args):
# Set logical unit number to pass between subroutines
lubfr = 11
luprt = 51
# Physical file to be opened
f = args
print '1'
# Open file for READING using FORTRAN open call
py_bufrlib.fortopen(lubfr,f,'R')
# Open file for WRITING using FORTRAN open call
py_bufrlib.fortopen(luprt,'dumpbufr.out','W')
print '2'
# Connect file to BUFRLIB library via openbf
py_bufrlib.openbf(11,'IN',11)
print '3'
# Set date length
py_bufrlib.datelen(10)
print '4'
# Attempt to read a subset from the BUFR file
ctr = 0
while True:
subset,jdate,iret = py_bufrlib.readns(11)
print 'SUBSET: ', subset, ', DATE: ', jdate, ', RET: ', iret, ' :: Count: ', ctr
py_bufrlib.ufbdmp(lubfr,luprt)
if iret == -1:
break
ctr=ctr+1
def __init__(self, args):
# Set logical unit number to pass between subroutines
lubfr = 11
luprt = 51
# Physical file to be opened
f = args
# Open file for READING using FORTRAN open call
py_bufrlib.fortopen(lubfr, f, 'R')
# Open file for WRITING using FORTRAN open call
py_bufrlib.fortopen(luprt, 'dumpbufr.out', 'W')
# Connect file to BUFRLIB library via openbf
py_bufrlib.openbf(lubfr, 'IN', lubfr)
py_bufrlib.datelen(10)
# Attempt to read a subset from the BUFR file
iret = 0
while iret != -1:
subset, jdate, iret = py_bufrlib.readns(lubfr)
print ' MESSAGE TYPE ' + subset + '\n\n'
lun, il, im = py_bufrlib.status(lubfr)
lunidx = lun - 1
ptag = stride_tricks.as_strided(py_bufrlib.tables.tag,\
strides=(py_bufrlib.tables.tag.shape[1],1))
ptyp = stride_tricks.as_strided(py_bufrlib.tables.typ,\
strides=(py_bufrlib.tables.typ.shape[1],1))
###########################################
# OK, in order to get character arrays FROM the BUFRLIB common block data
# we need to reorganize the arrays. Typically the N-dimensional character
# array returns from FORTRAN as
#
# ([dimenson length,]dimension length, character length)
#
# This, combined with FORTRAN ordering creates a messy way to get
# characters back into the expected string order
# See http://cens.ioc.ee/pipermail/f2py-users/2012-September/002320.html
# for an example.
#
# Fortunately, after stumbling around with Numpy strides and the help of this
# site: http://scipy-lectures.github.com/advanced/advanced_numpy/index.html#indexing-scheme-strides
# there seems to be a standardized way to reorder the array such that
# the resulting array is indexed as is expected when examining the FORTRAN code.
#
# So, for a "simple" character array like TYP(MAXJL) found in the BUFRLIB common
# block /TABLES/ where TYP is CHARACTER*3, it returns from BUFRLIB to python
# in the dimension (MAXJL,3). (Refer back to the first link for the example).
#
# We can rerder the array by using the following convention (essentially feeding
# the shape of the N+1 D array in reverse order so that the first stride element
# is the length of the character length):
#
# a = stride_tricks.as_strided(py_bufrlib."COMMONBLOCK"."VARIABLE",
# strides=(py_bufrlib."COMMONBLOCK"."VARIABLE".shape[N],\
# ...
# strides=(py_bufrlib."COMMONBLOCK"."VARIABLE".shape[1],1))
#
# We then access 'a' as we would a N-dimensional array, returning a
# numpy array with the characters in order. (use the .tostring() method to
# "string-ize" it if desired.)
#
# This has worked on a 1D character array (TAG, TYP in the above code) and a
# 2D character array (TABB, below). I have not tried this on anything greater
# than a 2D character array, though the logic is extensible, I assume.
#
###########################################
for nv in range(0, py_bufrlib.usrint.nval[lunidx]):
node = py_bufrlib.usrint.inv[nv, lunidx] - 1
nemo = ptag[node].tostring()
ityp = py_bufrlib.tables.itp[node]
type = ptyp[node].tostring()
numb = ''
desc = ''
unit = ''
rval = ''
# print '\tNODE: ', node
# print '\tNEMO: ' + nemo
# print '\tITYP: ', ityp
# print '\tTYPE: ', type
if ityp >= 1 and ityp <= 3:
idn, tab, nemtab_iret = py_bufrlib.nemtab(lun, nemo)
# print '\t\tIDN: ', idn
# print '\t\tTAB: >>' + tab + '<<'
# print '\t\tN: ', nemtab_iret
tabb = stride_tricks.as_strided(py_bufrlib.tababd.tabb,\
strides=(py_bufrlib.tababd.tabb.shape[2],\
py_bufrlib.tababd.tabb.shape[1],1))
numb = tabb[nemtab_iret - 1, lunidx][0:6].tostring()
desc = tabb[nemtab_iret - 1,
lunidx][15:69].tostring().rstrip()
unit = tabb[nemtab_iret - 1,
lunidx][70:93].tostring().rstrip()
rval = py_bufrlib.usrint.val[nv, lunidx]
#
# print '\t\tNUMB: ', numb
# print '\t\tDESC: ', desc
# print '\t\tUNIT: ', unit
# print '\t\tRVAL: >>>', rval, '<<<'
if ityp == 0 or ityp == 1:
continue #print 'SKIPPING DUE TO TYPE: ', type
elif ityp == 2:
if py_bufrlib.ibfms(float(rval)) != 0:
rval = 'MISSING'
print '{0:6s} {1:10s} {2:>20s} {3:24s} {4:48s}'.format(\
numb,nemo,rval,unit,desc)
else:
# if isinstance(rval, basestring):
print '{0:6s} {1:10s} {2:>20s} {3:24s} {4:48s}'.format(\
numb,nemo,str(rval),unit,desc)
# else:
# print '{0:6s} {1:10s} {2:20.2f} {3:24s} {4:48s}'.format(\
# numb,nemo,rval,unit,desc)
else:
continue
def __init__(self, args):
# Set logical unit number to pass between subroutines
lubfr = 11
luprt = 51
# Physical file to be opened
f = args
# Open file for READING using FORTRAN open call
py_bufrlib.fortopen(lubfr, f, "R")
# Open file for WRITING using FORTRAN open call
py_bufrlib.fortopen(luprt, "dumpbufr.out", "W")
# Connect file to BUFRLIB library via openbf
py_bufrlib.openbf(lubfr, "IN", lubfr)
py_bufrlib.datelen(10)
# Attempt to read a subset from the BUFR file
iret = 0
while iret != -1:
subset, jdate, iret = py_bufrlib.readns(lubfr)
print " MESSAGE TYPE " + subset + "\n\n"
lun, il, im = py_bufrlib.status(lubfr)
lunidx = lun - 1
ptag = stride_tricks.as_strided(py_bufrlib.tables.tag, strides=(py_bufrlib.tables.tag.shape[1], 1))
ptyp = stride_tricks.as_strided(py_bufrlib.tables.typ, strides=(py_bufrlib.tables.typ.shape[1], 1))
###########################################
# OK, in order to get character arrays FROM the BUFRLIB common block data
# we need to reorganize the arrays. Typically the N-dimensional character
# array returns from FORTRAN as
#
# ([dimenson length,]dimension length, character length)
#
# This, combined with FORTRAN ordering creates a messy way to get
# characters back into the expected string order
# See http://cens.ioc.ee/pipermail/f2py-users/2012-September/002320.html
# for an example.
#
# Fortunately, after stumbling around with Numpy strides and the help of this
# site: http://scipy-lectures.github.com/advanced/advanced_numpy/index.html#indexing-scheme-strides
# there seems to be a standardized way to reorder the array such that
# the resulting array is indexed as is expected when examining the FORTRAN code.
#
# So, for a "simple" character array like TYP(MAXJL) found in the BUFRLIB common
# block /TABLES/ where TYP is CHARACTER*3, it returns from BUFRLIB to python
# in the dimension (MAXJL,3). (Refer back to the first link for the example).
#
# We can rerder the array by using the following convention (essentially feeding
# the shape of the N+1 D array in reverse order so that the first stride element
# is the length of the character length):
#
# a = stride_tricks.as_strided(py_bufrlib."COMMONBLOCK"."VARIABLE",
# strides=(py_bufrlib."COMMONBLOCK"."VARIABLE".shape[N],\
# ...
# strides=(py_bufrlib."COMMONBLOCK"."VARIABLE".shape[1],1))
#
# We then access 'a' as we would a N-dimensional array, returning a
# numpy array with the characters in order. (use the .tostring() method to
# "string-ize" it if desired.)
#
# This has worked on a 1D character array (TAG, TYP in the above code) and a
# 2D character array (TABB, below). I have not tried this on anything greater
# than a 2D character array, though the logic is extensible, I assume.
#
###########################################
for nv in range(0, py_bufrlib.usrint.nval[lunidx]):
node = py_bufrlib.usrint.inv[nv, lunidx] - 1
nemo = ptag[node].tostring()
ityp = py_bufrlib.tables.itp[node]
type = ptyp[node].tostring()
numb = ""
desc = ""
unit = ""
rval = ""
# print '\tNODE: ', node
# print '\tNEMO: ' + nemo
# print '\tITYP: ', ityp
# print '\tTYPE: ', type
if ityp >= 1 and ityp <= 3:
idn, tab, nemtab_iret = py_bufrlib.nemtab(lun, nemo)
# print '\t\tIDN: ', idn
# print '\t\tTAB: >>' + tab + '<<'
# print '\t\tN: ', nemtab_iret
tabb = stride_tricks.as_strided(
py_bufrlib.tababd.tabb,
strides=(py_bufrlib.tababd.tabb.shape[2], py_bufrlib.tababd.tabb.shape[1], 1),
)
numb = tabb[nemtab_iret - 1, lunidx][0:6].tostring()
desc = tabb[nemtab_iret - 1, lunidx][15:69].tostring().rstrip()
unit = tabb[nemtab_iret - 1, lunidx][70:93].tostring().rstrip()
rval = py_bufrlib.usrint.val[nv, lunidx]
#
# print '\t\tNUMB: ', numb
# print '\t\tDESC: ', desc
# print '\t\tUNIT: ', unit
# print '\t\tRVAL: >>>', rval, '<<<'
if ityp == 0 or ityp == 1:
continue # print 'SKIPPING DUE TO TYPE: ', type
elif ityp == 2:
if py_bufrlib.ibfms(float(rval)) != 0:
rval = "MISSING"
print "{0:6s} {1:10s} {2:>20s} {3:24s} {4:48s}".format(numb, nemo, rval, unit, desc)
else:
# if isinstance(rval, basestring):
print "{0:6s} {1:10s} {2:>20s} {3:24s} {4:48s}".format(
numb, nemo, str(rval), unit, desc
)
# else:
# print '{0:6s} {1:10s} {2:20.2f} {3:24s} {4:48s}'.format(\
# numb,nemo,rval,unit,desc)
else:
continue
