def get_rec_dtype(em):
bad_fields = []
# Check if already have data type information for this class.
# Create it otherwise.
if em.class_ in dtype_table:
dtype = dtype_table[em.class_]
else:
print 'Creating entries for class:', obj.class_
fielddict = moose__.getFieldDict(obj.class_, 'valueFinfo')
print fielddict
keys = sorted(list(fielddict.keys()))
fields = [] # [('path', 'S1024')]
for fieldname in keys:
ftype = fielddict[fieldname]
# If we do not have the type of this field in cpp-np data
# type map, skip it. We can handle vectors as nested
# table, but the problem with that is numpy requires all
# entries to have a fixed size for this table. Since
# vectors can have arbitrary length, we cannot get such a
# fixed size without checking the size of the vector field
# in all elements.
if ftype in cpptonp:
fields.append((fieldname, cpptonp[ftype]))
dtype_table[obj.class_] = np.dtype(fields)
return dtype_table[obj.class_]
def get_rec_dtype(em):
bad_fields = []
# Check if already have data type information for this class.
# Create it otherwise.
if em.className in dtype_table:
dtype = dtype_table[em.className]
else:
print('Creating entries for class:', em.className)
fielddict = moose.getFieldDict(em.className, 'valueFinfo')
print(fielddict)
# If we do not have the type of this field in cpp-np data
# type map, skip it. We can handle vectors as nested
# table, but the problem with that is numpy requires all
# entries to have a fixed size for this table. Since
# vectors can have arbitrary length, we cannot get such a
# fixed size without checking the size of the vector field
# in all elements.
fields = [
(fieldname, cpptonp[ftype]) # [('path', 'S1024')]
for fieldname, ftype in sorted(fielddict.items())
if ftype in cpptonp
]
dtype_table[em.className] = np.dtype(fields)
return dtype_table[em.className]
def test_other():
a1 = moose.Pool('/ada')
assert a1.className == 'Pool', a1.className
finfo = moose.getFieldDict(a1.className)
s = moose.Streamer('/asdada')
p = moose.PulseGen('pg1')
assert p.delay[0] == 0.0
p.delay[1] = 0.99
assert p.delay[1] == 0.99, p.delay[1]
def __init__(self, mooseObject, parent=None):
"""Set up the model.
The table model has one moose field in each row. A field that
has a set method is editable. Fields listed in extra_fields
are not shown.
Members:
fields -- list of the names of the fields in the object.
plotNames -- lists the names of the available plot
windows. These are displayed as the targets in
the plot submenu / combobox.
fieldFlags -- flags for each field. We calculate these ahead
of time by checking if the field can be set, if
it is a numerical (so can be dragged to a plot
window).
"""
QtCore.QAbstractTableModel.__init__(self, parent)
self.mooseObject = mooseObject
self._header = ('Field', 'Value')
self.fields = []
self.plotNames = ['None']
self.fieldFlags = {}
self.fieldPlotNameMap = {}
try:
self.mooseObject = moose.element(mooseObject.getId())
except AttributeError:
config.LOGGER.error('Could not wrap object %s into class %s' % (mooseObject.path, className))
return
for fieldName in self.mooseObject.getFieldNames('valueFinfo'):
#print fieldName
if(fieldName in ObjectFieldsModel.extra_fields):
continue
else:
value = self.mooseObject.getField(fieldName)
#print value
flag = Qt.ItemIsEnabled | Qt.ItemIsSelectable
srchField = 'set_'+fieldName
try:
for fn in (fn for fn in moose.getFieldDict(self.mooseObject.class_,'destFinfo').keys() if fn.startswith(srchField)):
flag = flag | Qt.ItemIsEditable
value = mooseObject.getField(fieldName)
except Exception, e:
config.LOGGER.error("%s" % (e))
self.fieldFlags[fieldName] = flag
self.fields.append(fieldName)
def test_other():
a1 = moose.Pool('/ada')
assert a1.className == 'Pool', a1.className
finfo = moose.getFieldDict(a1.className)
s = moose.Streamer('/asdada')
p = moose.PulseGen('pg1')
assert p.delay[0] == 0.0
p.delay[1] = 0.99
assert p.delay[1] == 0.99, p.delay[1]
c = moose.Stoich('/dadaa')
v1 = moose.getFieldNames(c)
v2 = c.getFieldNames()
assert v1 == v2
assert len(v1) > 10, v1
def setSelectedElements(self, elementlist):
"""Create a grid of widgets displaying paths of elements in
`elementlist` if it has at least one plottable field (a field
with a numeric value). The numeric fields are listed in a
combobox next to the element path and can be selected for
plotting by the user.
"""
for ii in range(self.getPlotListWidget().layout().count()):
item = self.getPlotListWidget().layout().itemAt(ii)
if item is None:
continue
self.getPlotListWidget().layout().removeItem(item)
w = item.widget()
w.hide()
del w
del item
self._elementWidgetsDict.clear()
label = QLabel('Element')
label.setSizePolicy(QSizePolicy.Fixed, QSizePolicy.Fixed)
self.getPlotListWidget().layout().addWidget(label, 0, 0, 1, 2)
self.getPlotListWidget().layout().addWidget(QLabel('Fields to plot'),
0, 2, 1, 1)
for ii, entry in enumerate(elementlist):
el = moose.element(entry)
plottableFields = []
for field, dtype in list(
moose.getFieldDict(el.className, 'valueFinfo').items()):
if dtype == 'double':
plottableFields.append(field)
if len(plottableFields) == 0:
continue
elementLabel = QLabel(el.path)
fieldsCombo = CheckComboBox(self)
fieldsCombo.addItem('')
for item in plottableFields:
fieldsCombo.addItem(item)
self.getPlotListWidget().layout().addWidget(
elementLabel, ii + 1, 0, 1, 2)
self.getPlotListWidget().layout().addWidget(
fieldsCombo, ii + 1, 2, 1, 1)
self._elementWidgetsDict[el] = (elementLabel, fieldsCombo)
def setSelectedElements(self, elementlist):
"""Create a grid of widgets displaying paths of elements in
`elementlist` if it has at least one plottable field (a field
with a numeric value). The numeric fields are listed in a
combobox next to the element path and can be selected for
plotting by the user.
"""
for ii in range(self.getPlotListWidget().layout().count()):
item = self.getPlotListWidget().layout().itemAt(ii)
if item is None:
continue
self.getPlotListWidget().layout().removeItem(item)
w = item.widget()
w.hide()
del w
del item
self._elementWidgetsDict.clear()
label = QtGui.QLabel('Element')
label.setSizePolicy(QtGui.QSizePolicy.Fixed, QtGui.QSizePolicy.Fixed)
self.getPlotListWidget().layout().addWidget(label, 0, 0, 1, 2)
self.getPlotListWidget().layout().addWidget(QtGui.QLabel('Fields to plot'), 0, 2, 1, 1)
for ii, entry in enumerate(elementlist):
el = moose.element(entry)
plottableFields = []
for field, dtype in list(moose.getFieldDict(el.className, 'valueFinfo').items()):
if dtype == 'double':
plottableFields.append(field)
if len(plottableFields) == 0:
continue
elementLabel = QtGui.QLabel(el.path)
fieldsCombo = CheckComboBox(self)
fieldsCombo.addItem('')
for item in plottableFields:
fieldsCombo.addItem(item)
self.getPlotListWidget().layout().addWidget(elementLabel, ii+1, 0, 1, 2)
self.getPlotListWidget().layout().addWidget(fieldsCombo, ii+1, 2, 1, 1)
self._elementWidgetsDict[el] = (elementLabel, fieldsCombo)
def get_rec_dtype(em):
bad_fields = []
# Check if already have data type information for this class.
# Create it otherwise.
if em.className in dtype_table:
dtype = dtype_table[em.className]
else:
print('Creating entries for class:', em.className)
fielddict = moose.getFieldDict(em.className, 'valueFinfo')
print(fielddict)
# If we do not have the type of this field in cpp-np data
# type map, skip it. We can handle vectors as nested
# table, but the problem with that is numpy requires all
# entries to have a fixed size for this table. Since
# vectors can have arbitrary length, we cannot get such a
# fixed size without checking the size of the vector field
# in all elements.
fields = [(fieldname, cpptonp[ftype]) # [('path', 'S1024')]
for fieldname, ftype in sorted(fielddict.items())
if ftype in cpptonp]
dtype_table[em.className] = np.dtype(fields)
return dtype_table[em.className]
def __init__(self, datain, headerdata=['Field','Value'], parent=None, *args):
QAbstractTableModel.__init__(self, parent, *args)
self.fieldFlags = {}
self.fields = []
self.mooseObject = datain
self.headerdata = headerdata
for fieldName in self.mooseObject.getFieldNames('valueFinfo'):
if(fieldName in extra_fields):
continue
else:
value = self.mooseObject.getField(fieldName)
self.fields.append(fieldName)
flag = Qt.ItemIsEnabled | Qt.ItemIsSelectable
srchField = 'set_'+fieldName
try:
for fn in (fn for fn in moose.getFieldDict(self.mooseObject.class_,'destFinfo').keys() if fn.startswith(srchField)):
flag = flag | Qt.ItemIsEditable
value = self.mooseObject.getField(fieldName)
except Exception, e:
pass
self.fieldFlags[fieldName] = flag
def savestate(filename=None):
"""Dump the state of MOOSE in an hdf5 file.
The file will have a data set for each class.
Each such dataset will be a column of field values.
This function needs more work on moose serialization.
"""
if filename is None:
filename = 'moose_session_' + time.strftime('%Y%m%d_%H%M%S') + '.hdf5'
with h5.File(filename, 'w') as fd:
root = fd.create_group('/elements')
meta = fd.create_group('/metadata')
typeinfo_dataset = meta.create_dataset('typeinfo', shape=(size_step,), dtype=[('path', 'S1024'), ('class', 'S64'), ('dims', 'S64'), ('parent', 'S1024')], compression='gzip', compression_opts=6)
typeinfo = []
class_dataset_dict = {}
class_count_dict = {}
class_array_dict = {}
objcount = 0
for obj in moose__.wildcardFind("/##"):
if obj.path.startswith('/Msg') or obj.path.startswith('/class') or obj.class_ == 'Table' or obj.class_ == 'TableEntry':
continue
print 'Processing:', obj.path, obj.class_
typeinfo.append((obj.path, obj.class_, str(obj.shape), obj[0].parent.path))
objcount += 1
if len(typeinfo) == size_step:
typeinfo_dataset.resize(objcount)
typeinfo_dataset[objcount - size_step: objcount] = np.rec.array(typeinfo, typeinfo_dataset.dtype)
typeinfo = []
# If we do not yet have dataset for this class, create one and keep it in dict
if obj.class_ not in class_dataset_dict:
print 'Creating entries for class:', obj.class_
fielddict = moose__.getFieldDict(obj.class_, 'valueFinfo')
print fielddict
keys = sorted(list(fielddict.keys()))
fields = [] # [('path', 'S1024')]
for fieldname in keys:
ftype = fielddict[fieldname]
if ftype in cpptonp:
fields.append((fieldname, cpptonp[ftype]))
elif ftype == 'Id' or ftype == 'ObjId':
fields.append((fieldname, 'S1024'))
# print fields
ds = root.create_dataset(obj.class_, shape=(size_step,), dtype=fields, compression='gzip', compression_opts=6)
class_dataset_dict[obj.class_] = ds
class_array_dict[obj.class_] = []
class_count_dict[obj.class_] = 0
# Lookup the dataset for the class this object belongs to
ds = class_dataset_dict[obj.class_]
for entry in obj:
fields = []
print entry.path,
for f in ds.dtype.names:
print 'getting field:', f
entry.getField(f)
fields = [f.path if isinstance(f, moose__.ematrix) or isinstance(f, moose__.element) else f for f in fields]
class_array_dict[obj.class_].append(fields)
# print 'fields:'
# print fields
# print 'length:', len(class_array_dict[obj.class_])
class_count_dict[obj.class_] += 1
if class_count_dict[obj.class_] == size_step:
oldlen = ds.len()
if oldlen <= class_count_dict[obj.class_]:
ds.resize((class_count_dict[obj.class_]))
ds[oldlen: class_count_dict[obj.class_]] = np.rec.array(class_array_dict[obj.class_], dtype=ds.dtype)
class_array_dict[obj.class_] = []
for classname in class_array_dict:
ds = class_dataset_dict[classname]
ds.resize((class_count_dict[classname], ))
if len(class_array_dict[classname]) > 0:
start = class_count_dict[classname] - len(class_array_dict[classname])
ds[start:] = np.rec.array(class_array_dict[classname], dtype=ds.dtype)
if len(typeinfo) > 0:
typeinfo_dataset.resize((objcount,))
typeinfo_dataset[objcount-len(typeinfo): objcount] = np.rec.array(typeinfo, dtype=typeinfo_dataset.dtype)
def savestate(filename=None):
"""Dump the state of MOOSE in an hdf5 file.
The file will have a data set for each class.
Each such dataset will be a column of field values.
This function needs more work on moose serialization.
"""
if filename is None:
filename = 'moose_session_' + time.strftime('%Y%m%d_%H%M%S') + '.hdf5'
with h5.File(filename, 'w') as fd:
root = fd.create_group('/elements')
meta = fd.create_group('/metadata')
typeinfo_dataset = meta.create_dataset('typeinfo',
shape=(size_step, ),
dtype=[('path', 'S1024'),
('class', 'S64'),
('dims', 'S64'),
('parent', 'S1024')],
compression='gzip',
compression_opts=6)
typeinfo = []
class_dataset_dict = {}
class_count_dict = {}
class_array_dict = {}
objcount = 0
for obj in moose.wildcardFind("/##"):
if obj.path.startswith('/Msg') or obj.path.startswith(
'/class'
) or obj.className == 'Table' or obj.className == 'TableEntry':
continue
print('Processing:', obj.path, obj.className)
typeinfo.append(
(obj.path, obj.className, str(obj.shape), obj[0].parent.path))
objcount += 1
if len(typeinfo) == size_step:
typeinfo_dataset.resize(objcount)
typeinfo_dataset[objcount - size_step:objcount] = np.rec.array(
typeinfo, typeinfo_dataset.dtype)
typeinfo = []
# If we do not yet have dataset for this class, create one and keep it in dict
if obj.className not in class_dataset_dict:
print('Creating entries for class:', obj.className)
fielddict = moose.getFieldDict(obj.className, 'valueFinfo')
print(fielddict)
keys = sorted(fielddict)
fields = [] # [('path', 'S1024')]
for fieldname in keys:
ftype = fielddict[fieldname]
if ftype in cpptonp:
fields.append((fieldname, cpptonp[ftype]))
elif ftype == 'Id' or ftype == 'ObjId':
fields.append((fieldname, 'S1024'))
# print fields
ds = root.create_dataset(obj.className,
shape=(size_step, ),
dtype=fields,
compression='gzip',
compression_opts=6)
class_dataset_dict[obj.className] = ds
class_array_dict[obj.className] = []
class_count_dict[obj.className] = 0
# Lookup the dataset for the class this object belongs to
ds = class_dataset_dict[obj.className]
for entry in obj:
fields = []
print(entry.path, end=' ')
for f in ds.dtype.names:
print('getting field:', f)
entry.getField(f)
fields = [
f.path if isinstance(f, moose.vec)
or isinstance(f, moose.element) else f for f in fields
]
class_array_dict[obj.className].append(fields)
# print 'fields:'
# print fields
# print 'length:', len(class_array_dict[obj.className])
class_count_dict[obj.className] += 1
if class_count_dict[obj.className] == size_step:
oldlen = ds.len()
if oldlen <= class_count_dict[obj.className]:
ds.resize((class_count_dict[obj.className]))
ds[oldlen:class_count_dict[obj.className]] = np.rec.array(
class_array_dict[obj.className], dtype=ds.dtype)
class_array_dict[obj.className] = []
for classname in class_array_dict:
ds = class_dataset_dict[classname]
ds.resize((class_count_dict[classname], ))
if len(class_array_dict[classname]) > 0:
start = class_count_dict[classname] - len(
class_array_dict[classname])
ds[start:] = np.rec.array(class_array_dict[classname],
dtype=ds.dtype)
if len(typeinfo) > 0:
typeinfo_dataset.resize((objcount, ))
typeinfo_dataset[objcount - len(typeinfo):objcount] = np.rec.array(
typeinfo, dtype=typeinfo_dataset.dtype)
