method Field.toVTK2 not working for tensors? about mupif HOT 7 CLOSED

Hi Grauchs,
I enhanced field2VTKData() and field2Image2D() for Tensors. The example is under Example10-stacTM-local/TensorFieldTest.py. I checked generated VTK files in paraview and it works. Note that tensors need 3x3 list in each vertex (cell), which is what expects pyvtk module at the end.

from mupif.

When I run TensorFieldTest.py from Example 10, I get the following error message:
Traceback (most recent call last):
File "TensorFieldTest.py", line 17, in
g.field2VTKData().tofile('Tensors')
File "/usr/local/lib/python3.6/site-packages/pyvtk/init.py", line 203, in tofile
f.write(self.to_string(format))
File "/usr/local/lib/python3.6/site-packages/pyvtk/init.py", line 187, in to_string
ret.append(self.point_data.to_string(format))
File "/usr/local/lib/python3.6/site-packages/pyvtk/Data.py", line 46, in to_string
ret += [a.to_string(format) for a in self.data]
File "/usr/local/lib/python3.6/site-packages/pyvtk/Data.py", line 46, in
ret += [a.to_string(format) for a in self.data]
File "/usr/local/lib/python3.6/site-packages/pyvtk/Tensors.py", line 29, in to_string
return '\n'.join(ret)
TypeError: sequence item 1: expected str instance, bytes found

I work under Linux Centos6, with python3.6, pyvtk0.5.18 and mupif 1.1.6.
Could the problem be related to this issue:
https://stackoverflow.com/questions/33054527/python-3-5-typeerror-a-bytes-like-object-is-required-not-str-when-writing-t

from mupif.

Dear Grauchs,
this problem is related to pyvtk module under Python3 and partially missing conversion from strings to bytes. Python2 works OK. I submitted this bug&fix to developers, however, you can fix it easily by modifying Tensors.py (mine is /usr/local/lib/python3.5/dist-packages/pyvtk/Tensors.py). Add bold text so it looks like:

def to_string(self,format='ascii'):
t = self.get_datatype(self.tensors)
ret = [('TENSORS %s %s'%(self.name,t)).encode(),
self.seq_to_string(self.tensors,format,t)]
return b'\n'.join(ret)

Regards. Vit.

from mupif.

After switching from mupif 1 to the most recent version of mupif version 2, I got the following warnings and error messages when using Field.toVTK2 with tensors. Note that for scalars and vectors the field output with Field.2toVTK2 works flawlessly. There seems to be again an issue with the tensor format 3 x 3 values versus 1 * 9 values. For the error message, I have got no real clue! Could it be related to the unit I give to getField?

2017-12-11 09:47:51 WARNING:common.py:178 expected 3-sequence but got 9-<class 'tuple'>

2017-12-11 09:47:51 WARNING:common.py:178 expected 3-sequence but got 9-<class 'tuple'>

2017-12-11 09:47:51 WARNING:common.py:178 expected 3-sequence but got 9-<class 'tuple'>

Traceback (most recent call last):
File "testrun-v4.py", line 66, in
res3.toVTK2('testoutput-s',format='ascii')
File "/usr/local/lib/python3.6/site-packages/mupif-1.1.6-py3.6.egg/mupif/Field.py", line 703, in toVTK2
self.field2VTKData().tofile(filename=fileName,format=format)
File "/usr/local/lib/python3.6/site-packages/mupif-1.1.6-py3.6.egg/mupif/Field.py", line 406, in field2VTKData
pyvtk.PointData(pyvtk.Tensors(self.value,**vectorsKw),lookupTable),'Unstructured Grid Example')
File "/usr/local/lib/python3.6/site-packages/pyvtk/init.py", line 167, in init
raise ValueError('DataSet (size=%s) and PointData (size=%s) have different sizes'%(s,s1))
ValueError: DataSet (size=60) and PointData (size=1) have different sizes

from mupif.

Hi,
tensors should be fixed now in dev-2.0 version. If you wait a few more days, dev-2.0 and master will be merged together and there will be master git version only. (The tensor problem emerged due to separate branches, merging would solve it later).

from mupif.

The problem is now solved!

from mupif.

I installed mupif-2.0.0. Now, all problems which reappeared when I switched to mupif-dev-2.0 have disappeared, as you predicted.

from mupif.