def test_event_offset():
from numpy import asarray, rec
data = [1, 2, 3]
assert_array_equal(util.event_offset(data, 1), [2, 3, 4])
assert_array_equal(util.event_offset(asarray(data), 2), [3, 4, 5])
marked = rec.fromarrays((data, ['a', 'b', 'c']), names=('start', 'names'))
assert_array_equal(util.event_offset(marked, 1)['start'], [2, 3, 4])
def recarray_from_file(source, ifo=None, columns=None, loudest=False):
"""Read a `GWRecArray` from a PyCBC live HDF5 file
"""
# read HDF5 file
if isinstance(source, CacheEntry):
source = source.path
if isinstance(source, str):
h5f = source = h5py.File(source, 'r')
opened = True
else:
opened = False
# find group
if isinstance(source, h5py.File):
if ifo is None:
try:
ifo, = list(source)
except ValueError as e:
e.args = ("PyCBC live HDF5 file contains multiple IFO groups, "
"please select ifo manually", )
raise
try:
source = source[ifo]
except KeyError as e:
e.args = ("No group for ifo %r in PyCBC live HDF5 file" % ifo, )
raise
# at this stage, 'source' should be an HDF5 group in the pycbc live format
if columns is None:
columns = [c for c in source if c not in INVALID_COLUMNS]
names, data = zip(*[(k, source[k][:]) for k in source if k in columns])
names = list(map(str, names))
if loudest: # recover only the 'loudest' events
loudest = source['loudest'][:]
data = [d[loudest] for d in data]
else:
data = list(data)
# calculate new_snr on-the-fly
if 'new_snr' in columns and 'new_snr' not in source:
# get columns needed for newsnr
snr = data[names.index('snr')]
rchisq = data[names.index('chisq')] # chisq is already reduced
# calculate and append to column list
data.append(get_new_snr(snr, rchisq))
names.append('new_snr')
# calculate mchirp
if 'mchirp' in columns and 'mchirp' not in source:
mass1 = data[names.index('mass1')]
mass2 = data[names.index('mass2')]
data.append(get_mchirp(mass1, mass2))
names.append('mchirp')
# read columns into numpy recarray
out = rec.fromarrays(data, names=map(str, names)).view(GWRecArray)
if 'end_time' in columns:
out.sort(order='end_time')
if opened:
h5f.close()
return out
def recarray_from_file(source, ifo=None, columns=None, loudest=False):
"""Read a `GWRecArray` from a PyCBC live HDF5 file
"""
# read HDF5 file
if isinstance(source, CacheEntry):
source = source.path
if isinstance(source, str):
h5f = source = h5py.File(source, 'r')
opened = True
else:
opened = False
# find group
if isinstance(source, h5py.File):
if ifo is None:
try:
ifo, = list(source)
except ValueError as e:
e.args = ("PyCBC live HDF5 file contains multiple IFO groups, "
"please select ifo manually",)
raise
try:
source = source[ifo]
except KeyError as e:
e.args = ("No group for ifo %r in PyCBC live HDF5 file" % ifo,)
raise
# at this stage, 'source' should be an HDF5 group in the pycbc live format
if columns is None:
columns = [c for c in source if c not in INVALID_COLUMNS]
names, data = zip(*[(k, source[k][:]) for k in source if k in columns])
names = list(map(str, names))
if loudest: # recover only the 'loudest' events
loudest = source['loudest'][:]
data = [d[loudest] for d in data]
else:
data = list(data)
# calculate new_snr on-the-fly
if 'new_snr' in columns and 'new_snr' not in source:
# get columns needed for newsnr
snr = data[names.index('snr')]
rchisq = data[names.index('chisq')] # chisq is already reduced
# calculate and append to column list
data.append(get_new_snr(snr, rchisq))
names.append('new_snr')
# calculate mchirp
if 'mchirp' in columns and 'mchirp' not in source:
mass1 = data[names.index('mass1')]
mass2 = data[names.index('mass2')]
data.append(get_mchirp(mass1, mass2))
names.append('mchirp')
# read columns into numpy recarray
out = rec.fromarrays(data, names=map(str, names)).view(GWRecArray)
if 'end_time' in columns:
out.sort(order='end_time')
if opened:
h5f.close()
return out
def is_sorted(self):
idx_cols = list(self.schema.idx)
if len(idx_cols) == 1:
arr = self[idx_cols[0]]
return all(arr[1:] >= arr[:-1])
# Multi-column index we fallback on argsort
arr = rec.fromarrays([self[n] for n in idx_cols], names=idx_cols)
sort_mask = self.argsort()
a_range = arange(len(sort_mask))
return all(sort_mask == a_range)
def test_select(self):
f0 = arange(10, dtype=int32)
f1 = arange(10, dtype=float64)
irregular = rec.fromarrays([f0, f1])
f0 = irregular['f0']
f1 = irregular['f1']
i0 = evaluate('f0 < 5')
i1 = evaluate('f1 < 5')
assert_array_equal(f0[i0], arange(5, dtype=int32))
assert_array_equal(f1[i1], arange(5, dtype=float64))
def test_select(self):
f0 = arange(10, dtype=int32)
f1 = arange(10, dtype=float64)
irregular = rec.fromarrays([f0, f1])
f0 = irregular['f0']
f1 = irregular['f1']
i0 = evaluate('f0 < 5')
i1 = evaluate('f1 < 5')
assert_array_equal(f0[i0], arange(5, dtype=int32))
assert_array_equal(f1[i1], arange(5, dtype=float64))
def dict_list_to_frame(dict_list):
df = pd.DataFrame(dict_list)
d0 = dict( df.iloc[0] )
goodkeys = [ k for k in d0.keys() if (type(d0[k])!=fits.card.Undefined)]
df = df[goodkeys]
# comb = pdplus.df_to_rec_strings(df)
dfs = df.select_dtypes(include=['object'])
dfns = df.select_dtypes(exclude=['object'])
dfs = rec.fromarrays(np.array(dfs).astype('S100').T,names=list(dfs.columns))
names = list(dfns.columns)
arrs = [dfns[n] for n in names]
comb = mlab.rec_append_fields(dfs,names,arrs)
return comb
def extrema(x, max=True, min=True, withend=True):
"""
Return indexes, values, and sign of curvature of local extrema of 1-d array.
The boolean arguments max, min, withend determine whether to
include maxima and minima, and include the endpoints.
Basic usage.
>>> x = [2, 1, 0, 1, 2]
>>> extrema(x) # doctest: +NORMALIZE_WHITESPACE +ELLIPSIS
rec.array([(0, 2, -1), (2, 0, 1), (4, 2, -1)],
dtype=[('index', '<i...'), ('value', '<i...'), ('curv', '<i...')])
Options to include only certain types of extrema.
>>> extrema(x, withend=False)
rec.array([(2, 0, 1)],...
>>> extrema(x, max=False)
rec.array([(2, 0, 1)],...
>>> extrema(x, min=False)
rec.array([(0, 2, -1), (4, 2, -1)],...
>>> extrema(x, max=False, min=False)
rec.array([],...
The beginning and end of flat segments both count as extrema,
except the first and last data point.
>>> extrema([0, 0, 1, 1, 2, 2])
rec.array([(1, 0, 1), (2, 1, -1), (3, 1, 1), (4, 2, -1)],...
>>> extrema([0, 0, 0])
rec.array([],...)
>>> extrema([0, 0, 1, 1], withend=False)
rec.array([(1, 0, 1), (2, 1, -1)],...
@todo: Add options on how to handle flat segments.
"""
x = squeeze(x) # ensure 1-d numpy array
xpad = r_[x[1], x, x[-2]] # pad x so endpoints become minima or maxima
curv = sign(diff(sign(diff(xpad)))) # +1 at minima, -1 at maxima
i = curv.nonzero()[0] # nonzero() wraps the indices in a 1-tuple
ext = rec.fromarrays([i, x[i], curv[i]], names=["index", "value", "curv"])
if not withend: ext = ext[(i > 0) & (i < len(x) - 1)]
if not max: ext = ext[ext.curv >= 0]
if not min: ext = ext[ext.curv <= 0]
return ext
def torec(self):
"""
Returns a recarray, averaging across chains as needed.
"""
from numpy import sqrt, rec
fields = ['time', 'p.sd', 'p.mmse']
values = [self.tgrid, sqrt(_reduce(self.pvar)), _reduce(self.pmmse)]
if self.nchains > 1:
fields.append("p.mmse.sd")
values.append(self.pmmse.std(1))
for k, v in self.estimates.items():
fields.append(k)
values.append(_reduce(v))
if v.ndim > 1:
fields.append(k + ".sd")
values.append(v.std(1))
return rec.fromarrays(values, names=fields)
def load_recarray(group):
"""Read recarray from the given HDF5 group
"""
columns = map(str, list(group))
# read segments
try:
epoch = LIGOTimeGPS(group['segments'].attrs['epoch'])
except TypeError:
epoch = LIGOTimeGPS(float(group['segments'].attrs['epoch']))
segments = SegmentList(Segment(epoch + x[0], epoch + x[1]) for
x in group['segments'][:])
columns.pop(columns.index('segments'))
# read columns
data = [group[c] for c in columns]
# format and add
table = rec.fromarrays(data, names=columns).view(GWRecArray)
add_triggers(table, group.name.split('/')[-1], segments=segments)
return table
def argsort(self):
idx_cols = list(self.schema.idx)
arr = rec.fromarrays([self[n] for n in idx_cols], names=idx_cols)
# Mergesort is faster on pre-sorted arrays
return argsort(arr, kind="mergesort")
def reduce(self, *col_list, **col_dict):
"""
Return a new frame containing the choosen columns. A column can be
one of the existing column or an s-expression that we be automatically
evaluated.
"""
# Merge all args in one dict
columns = dict(zip(col_list, col_list))
columns.update(col_dict)
# Detect aggregations
all_ast = {}
for alias, expr in columns.items():
if expr.startswith("("):
all_ast[alias] = AST.parse(expr)
agg_ast = {}
other_ast = {}
for alias, ast in all_ast.items():
if ast.is_aggregate():
agg_ast[alias] = ast
else:
other_ast[alias] = ast
# Eval non-aggregated columns
env = self.eval_env()
non_agg = {}
for alias, expr in columns.items():
if alias in agg_ast:
continue
ast = other_ast.get(alias)
if ast:
arr = ast.eval(env=env)
else:
arr = self.columns[expr]
if isinstance(arr, Alias):
# un-pack alias
arr, alias = arr.value, arr.name
non_agg[alias] = arr
# Early exit if we don't need to compute aggregates
if not agg_ast:
schema = Schema.from_frame(non_agg, idx_columns=list(non_agg))
return Frame(schema, non_agg)
res = {}
if non_agg:
# Compute binning
records = rec.fromarrays(non_agg.values(), names=list(non_agg))
keys, bins = unique(records, return_inverse=True)
# Build resulting columns
for alias in non_agg:
arr = keys[alias]
if isinstance(arr, Alias):
# un-pack alias
arr, alias = arr.value, arr.name
res[alias] = arr
env.update({"_keys": keys, "_bins": bins})
# Compute aggregates
for alias, expr in agg_ast.items():
arr = expr.eval(env)
if isinstance(arr, Alias):
# un-pack alias
arr, alias = arr.value, arr.name
# Without bins, eval will return a scalar value
res[alias] = arr if non_agg else asarray([arr])
schema = Schema.from_frame(res, idx_columns=list(non_agg))
return Frame(schema, res)
from numpy import array, rec
from numpy.random import normal as nprandom
from rpy2.robjects import numpy2ri, r
foo = array(range(10))
bar = foo + nprandom(0,1,10)
d = rec.fromarrays([foo, bar], names=('foo','bar'))
print d
fit = r.lm('bar ~ foo', data=d)
print fit.rx2('coefficients')
def read(self, var, x=None, y=None, radius=0., tlim=None, ylim=None,
xlim=None, missions=None, sort=True, profile=True):
"""Reads dataset.
PARAMETERS
var (string) :
Variable to be read from dataset. It also accepts
special naming conventions in order to rename the
original dataset variable and to load alternative
variables in case of invalid data according to the
syntax '[new_var_name]:var[|other_var]'.
x, y (array like, optional) :
List of zonal and meridional point coordinate of
interest.
radius (float, optional) :
Search radius in degrees.
tlim, ylim, xlim (array like, optional) :
The temporal, meridional and zonal limits (minimum,
maximum) for which data will be read.
missions (array like, optional) :
List of missions to read data from. If omitted, defaults
available missions on dataset class intialization.
sort (boolean optional) :
If true, sorts the data record in order of ascendant
time, latitude and longitude.
profile (boolean, optional) :
Sets whether the status is send to screen.
RETURNS
dat (record array) :
Record time-series of 'time', 'latitude', 'longitude',
selected variable and 'mission'.
"""
t0 = time()
# Checks input parameters.
T = self.variables['time'].data
if var.find(':') >= 0: # Checks spetial variable syntax
var_name, var = var.split(':')
else:
var_name = var
if tlim == None:
tlim = (T.min(), T.max())
if (x != None) | (y != None):
x, y = asarray(x), asarray(y)
if x.size != y.size:
raise ValueError('Zonal and meridional coordinate dimensions '
'do not match.')
npoints = x.size
radius2 = radius ** 2
else:
npoints = 0
x = y = []
#
if ylim == None:
ylim = (-90., 90.)
if xlim == None:
xlim = (0., 360.)
else:
# Make sure longitude limits are between 0 and 360.
xlim = list(lon360(asarray(xlim)))
if missions == None:
missions = self.params['missions']
# First we have to select which files will be loaded, which will
# depend on the temporal limits given in $t$.
sel_time = flatnonzero((T >= floor(min(tlim))) &
(T <= ceil(max(tlim))))
N = len(sel_time)
# Second we will walk through each of the selected time in the dataset
# and load the correspondant file for the available missions.
t1 = time()
if profile:
s = '\rLoading data...'
stdout.write(s)
stdout.flush()
# Reset important variables
TIME, LAT, LON, VAR, MISSION = [array([])] * 5
#
for i, tm in enumerate(T[sel_time]):
t2 = time()
for (mission, dset, fname, cycle,
orbit) in self.attributes['time_dataset'][tm]:
# Skips mission not in missions list.
if mission not in missions:
continue
# Uncompresses gzipped file and opens NetCDF instance.
data = self.read_file('%s/%s/%s' % (self.params['path'],
mission, fname))
# Reads variable from NetCDF file.
raw_time = self.read_variable(data, 'time')
raw_lat = self.read_variable(data, 'lat')
raw_lon = self.read_variable(data, 'lon')
raw_dat = self.read_variable(data, var)
# Select relevant data range according to limit parameters
sel_from_time = (
(raw_time >= min(tlim)) & (raw_time <= max(tlim))
)
if (ylim != None) | (xlim !=None):
sel_from_limits = ones(data.dimensions['time'], dtype=bool)
else:
sel_from_limits = zeros(data.dimensions['time'],
dtype=bool)
if ylim != None:
sel_from_limits = (sel_from_limits &
((raw_lat >= min(ylim)) & (raw_lat <= max(ylim))))
if xlim != None:
sel_from_limits = (sel_from_limits &
((raw_lon >= min(xlim)) & (raw_lon <= max(xlim))))
# Select relevant data according to points and search radius.
sel_from_radius = zeros(data.dimensions['time'], dtype=bool)
for xx, yy in zip(x, y):
distance2 = ((raw_lat - yy) ** 2 +
(raw_lon - lon360(xx)) ** 2)
sel_from_radius = sel_from_radius | (distance2 <= radius2)
#
sel_data = flatnonzero(sel_from_time &
(sel_from_limits | sel_from_radius) & (~isnan(raw_dat)))
_time = raw_time[sel_data]
_lat = raw_lat[sel_data]
_lon = raw_lon[sel_data]
_dat = raw_dat[sel_data]
#
TIME = append(TIME, _time)
LAT = append(LAT, _lat)
LON = append(LON, _lon)
VAR = append(VAR, _dat)
MISSION = append(MISSION, [mission] * len(sel_data))
#
self.close_file(data)
#
# Profiling
if profile:
s = '\rLoading data... %s ' % (profiler(N, i+1, t0, t1, t2),)
stdout.write(s)
stdout.flush()
#
if profile:
stdout.write('\n')
stdout.flush()
# Converts the data a structured array
DAT = rec.fromarrays((TIME, LAT, LON, VAR, MISSION),
dtype=[('time', float64), ('latitude', float64),
('longitude', float64), (var_name, float64), ('mission', '|S3')])
# Some data sorting?
if sort:
DAT.sort(order=('time', 'latitude', 'longitude'), axis=0)
return DAT
def read(self,
var,
x=None,
y=None,
radius=0.,
tlim=None,
ylim=None,
xlim=None,
missions=None,
sort=True,
profile=True):
"""Reads dataset.
PARAMETERS
var (string) :
Variable to be read from dataset. It also accepts
special naming conventions in order to rename the
original dataset variable and to load alternative
variables in case of invalid data according to the
syntax '[new_var_name]:var[|other_var]'.
x, y (array like, optional) :
List of zonal and meridional point coordinate of
interest.
radius (float, optional) :
Search radius in degrees.
tlim, ylim, xlim (array like, optional) :
The temporal, meridional and zonal limits (minimum,
maximum) for which data will be read.
missions (array like, optional) :
List of missions to read data from. If omitted, defaults
available missions on dataset class intialization.
sort (boolean optional) :
If true, sorts the data record in order of ascendant
time, latitude and longitude.
profile (boolean, optional) :
Sets whether the status is send to screen.
RETURNS
dat (record array) :
Record time-series of 'time', 'latitude', 'longitude',
selected variable and 'mission'.
"""
t0 = time()
# Checks input parameters.
T = self.variables['time'].data
if var.find(':') >= 0: # Checks spetial variable syntax
var_name, var = var.split(':')
else:
var_name = var
if tlim == None:
tlim = (T.min(), T.max())
if (x != None) | (y != None):
x, y = asarray(x), asarray(y)
if x.size != y.size:
raise ValueError('Zonal and meridional coordinate dimensions '
'do not match.')
npoints = x.size
radius2 = radius**2
else:
npoints = 0
x = y = []
#
if ylim == None:
ylim = (-90., 90.)
if xlim == None:
xlim = (0., 360.)
else:
# Make sure longitude limits are between 0 and 360.
xlim = list(lon360(asarray(xlim)))
if missions == None:
missions = self.params['missions']
# First we have to select which files will be loaded, which will
# depend on the temporal limits given in $t$.
sel_time = flatnonzero((T >= floor(min(tlim)))
& (T <= ceil(max(tlim))))
N = len(sel_time)
# Second we will walk through each of the selected time in the dataset
# and load the correspondant file for the available missions.
t1 = time()
if profile:
s = '\rLoading data...'
stdout.write(s)
stdout.flush()
# Reset important variables
TIME, LAT, LON, VAR, MISSION = [array([])] * 5
#
for i, tm in enumerate(T[sel_time]):
t2 = time()
for (mission, dset, fname, cycle,
orbit) in self.attributes['time_dataset'][tm]:
# Skips mission not in missions list.
if mission not in missions:
continue
# Uncompresses gzipped file and opens NetCDF instance.
data = self.read_file('%s/%s/%s' %
(self.params['path'], mission, fname))
# Reads variable from NetCDF file.
raw_time = self.read_variable(data, 'time')
raw_lat = self.read_variable(data, 'lat')
raw_lon = self.read_variable(data, 'lon')
raw_dat = self.read_variable(data, var)
# Select relevant data range according to limit parameters
sel_from_time = ((raw_time >= min(tlim)) &
(raw_time <= max(tlim)))
if (ylim != None) | (xlim != None):
sel_from_limits = ones(data.dimensions['time'], dtype=bool)
else:
sel_from_limits = zeros(data.dimensions['time'],
dtype=bool)
if ylim != None:
sel_from_limits = (sel_from_limits &
((raw_lat >= min(ylim)) &
(raw_lat <= max(ylim))))
if xlim != None:
sel_from_limits = (sel_from_limits &
((raw_lon >= min(xlim)) &
(raw_lon <= max(xlim))))
# Select relevant data according to points and search radius.
sel_from_radius = zeros(data.dimensions['time'], dtype=bool)
for xx, yy in zip(x, y):
distance2 = ((raw_lat - yy)**2 + (raw_lon - lon360(xx))**2)
sel_from_radius = sel_from_radius | (distance2 <= radius2)
#
sel_data = flatnonzero(sel_from_time
& (sel_from_limits | sel_from_radius)
& (~isnan(raw_dat)))
_time = raw_time[sel_data]
_lat = raw_lat[sel_data]
_lon = raw_lon[sel_data]
_dat = raw_dat[sel_data]
#
TIME = append(TIME, _time)
LAT = append(LAT, _lat)
LON = append(LON, _lon)
VAR = append(VAR, _dat)
MISSION = append(MISSION, [mission] * len(sel_data))
#
self.close_file(data)
#
# Profiling
if profile:
s = '\rLoading data... %s ' % (profiler(N, i + 1, t0, t1,
t2), )
stdout.write(s)
stdout.flush()
#
if profile:
stdout.write('\n')
stdout.flush()
# Converts the data a structured array
DAT = rec.fromarrays((TIME, LAT, LON, VAR, MISSION),
dtype=[('time', float64), ('latitude', float64),
('longitude', float64),
(var_name, float64), ('mission', '|S3')])
# Some data sorting?
if sort:
DAT.sort(order=('time', 'latitude', 'longitude'), axis=0)
return DAT
def read(self, x=None, y=None, radius=0., tlim=None, ylim=None, xlim=None,
missions=None, sort=True, profile=True):
"""Reads dataset.
PARAMETERS
x, y (array like, optional) :
List of zonal and meridional point coordinate of
interest.
radius (float, optional) :
Search radius in degrees.
tlim, ylim, xlim (array like, optional) :
The temporal, meridional and zonal limits (minimum,
maximum) for which data will be read.
missions (array like, optional) :
List of missions to read data from. If omitted, defaults
available missions on dataset class intialization.
sort (boolean optional) :
If true, sorts the data record in order of ascendant
time, latitude and longitude.
profile (boolean, optional) :
Sets whether the status is send to screen.
RETURNS
dat (record array) :
Record time-series of 'time', 'latitude', 'longitude',
selected variable and 'mission'.
"""
t0 = time()
# Checks input parameters.
T = self.variables['time'].data
if tlim == None:
tlim = (T.min(), T.max())
if (x != None) | (y != None):
x, y = asarray(x), asarray(y)
if x.size != y.size:
raise ValueError('Zonal and meridional coordinate dimensions '
'do not match.')
npoints = x.size
radius2 = radius ** 2
else:
npoints = 0
x = y = []
#
if ylim == None:
ylim = (-90., 90.)
if xlim == None:
xlim = (0., 360.)
else:
# Make sure longitude limits are between 0 and 360.
xlim = list(lon360(asarray(xlim)))
if missions == None:
missions = self.params['missions']
# Aviso uses time in days since 1950-01-01 00:00:00 UTC, therefore
# we have to calculate the initial time in matplotlib's format. We
# also have to determine the proper variable using product name.
T0 = dates.datestr2num('1950-01-01 00:00:00 UTC')
var = self.params['product'].upper()
# First we have to select which files will be loaded, which will
# depend on the temporal limits given in $t$.
sel_time = flatnonzero((T >= floor(min(tlim))) &
(T <= ceil(max(tlim))))
N = len(sel_time)
# Second we will walk through each of the selected time in the dataset
# and load the correspondant file for the available missions.
t1 = time()
if profile:
s = '\rLoading data...'
stdout.write(s)
stdout.flush()
# Reset important variables
TIME, LAT, LON, VAR, MISSION = [array([])] * 5
#
for i, tm in enumerate(T[sel_time]):
t2 = time()
for (mission, fname) in self.attributes['time_dataset'][tm]:
# Skips mission not in missions list.
if mission not in missions:
continue
# Uncompresses gzipped file and opens NetCDF instance.
data = self.read_file('%s/%s/%s' % (self.params['path'],
mission, fname))
# Retrieve the scale factor for each variable
scale_lat = data.variables['latitude'].scale_factor
scale_lon = data.variables['latitude'].scale_factor
scale_dat = data.variables[var].scale_factor
# Get the raw time, latitude and longitude
raw_time = data.variables['time'].data + T0
raw_lat = data.variables['latitude'].data * scale_lat
raw_lon = data.variables['longitude'].data * scale_lon
# Select relevant data range according to limit parameters
sel_from_time = (
(raw_time >= min(tlim)) & (raw_time <= max(tlim))
)
sel_from_limits = zeros(data.dimensions['time'], dtype=bool)
if ylim != None:
sel_from_limits = (sel_from_limits |
((raw_lat >= min(ylim)) & (raw_lat <= max(ylim))))
if xlim != None:
sel_from_limits = (sel_from_limits |
((raw_lon >= min(xlim)) & (raw_lon <= max(xlim))))
# Select relevant data according to points and search radius.
sel_from_radius = zeros(data.dimensions['time'], dtype=bool)
for xx, yy in zip(x, y):
distance2 = ((raw_lat - yy) ** 2 +
(raw_lon - lon360(xx)) ** 2)
sel_from_radius = sel_from_radius | (distance2 <= radius2)
#
sel_data = flatnonzero(sel_from_time &
(sel_from_limits | sel_from_radius))
_time = raw_time[sel_data]
_lat = raw_lat[sel_data]
_lon = raw_lon[sel_data]
_dat = data.variables[var].data[sel_data] * scale_dat
#
TIME = append(TIME, _time)
LAT = append(LAT, _lat)
LON = append(LON, _lon)
VAR = append(VAR, _dat)
MISSION = append(MISSION, [mission] * len(sel_data))
#
self.close_file(data)
#
# Profiling
if profile:
s = '\rLoading data... %s ' % (profiler(N, i+1, t0, t1, t2),)
stdout.write(s)
stdout.flush()
#
if profile:
stdout.write('\n')
stdout.flush()
# Converts the data a structured array
DAT = rec.fromarrays((TIME, LAT, LON, VAR, MISSION),
dtype=[('time', float64), ('latitude', float64),
('longitude', float64), (self.params['product'], float64),
('mission', '|S3')])
#DAT = hstack((TIME[:, None], LAT[:, None], LON[:, None],
# VAR[:, None], MISSION[:, None])).view(dtype=[('time', float64),
# ('latitude', float64), ('longitude', float64),
# (self.params['product'], float64), ('mission', '|S3')])
# Some data sorting?
if sort:
DAT.sort(order=('time', 'latitude', 'longitude'), axis=0)
return DAT