format.write([0, 1, None]) with format='(3I10)' runs into TypeError

[hyperkang]

>>> from fortranformat import FortranRecordWriter
>>> format = FortranRecordWriter('(3I10)')
>>> output = format.write([0, 1, None])
Traceback (most recent call last):
  File "<input>", line 1, in <module>
  File "D:\simright-dev\py-fortranformat\fortranformat\FortranRecordWriter.py", line 42, in write
    return _output(self._eds, self._rev_eds, values)
  File "D:\simright-dev\py-fortranformat\fortranformat\_output.py", line 93, in output
    sub_string = _compose_i_string(
  File "D:\simright-dev\py-fortranformat\fortranformat\_output.py", line 641, in _compose_i_string
    val = int(val)
TypeError: int() argument must be a string, a bytes-like object or a number, not 'NoneType'

[brendanarnold]

Hi, could you post what you expect the behaviour to be?

[hyperkang]

Hi, could you post what you expect the behaviour to be?

I expect it behaves like the example in wiki guide:

>>> from fortranformat import FortranRecordWriter
>>> format = FortranRecordWriter('(3I10)')
>>> output = format.write([0.0, 1.0, None])
>>> output
'         0         1'

Or it might be easier to write None value into spaces with specified width, that is:

FortranRecordWriter('(3I10)').write([0, 1, None]) => '         0         1          '
FortranRecordWriter('(3I10)').write([0, None, 1]) => '         0                   1'

[hyperkang]

BTW, I would expect real numbers behave the similar way, that is:

>>> FortranRecordWriter('(3E10.2)').write([1.0, 2.0, None])
'  0.10E+01  0.20E+01          '
>>> FortranRecordWriter('(3E10.2)').write([1.0, None, 2.0])
'  0.10E+01            0.20E+01'

However, currently the code behaves as follows:

FortranRecordWriter('(3E10.2)').write([1.0, 2.0, None])

>>> FortranRecordWriter('(3E10.2)').write([1.0, 2.0, None])
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/home/simright/.pyenv/versions/pypi_fortranformat/lib/python3.7/site-packages/fortranformat/FortranRecordWriter.py", line 42, in write
    return _output(self._eds, self._rev_eds, values)
  File "/home/simright/.pyenv/versions/pypi_fortranformat/lib/python3.7/site-packages/fortranformat/_output.py", line 116, in output
    sub_string = _compose_float_string(w, e, d, state, val, 'E')
  File "/home/simright/.pyenv/versions/pypi_fortranformat/lib/python3.7/site-packages/fortranformat/_output.py", line 253, in _compose_float_string
    sign_bit = val < 0
TypeError: '<' not supported between instances of 'NoneType' and 'int'

FortranRecordWriter('(3E10.2)').write([1.0, None, 2.0])

>>> FortranRecordWriter('(3E10.2)').write([1.0, None, 2.0])
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/home/simright/.pyenv/versions/pypi_fortranformat/lib/python3.7/site-packages/fortranformat/FortranRecordWriter.py", line 42, in write
    return _output(self._eds, self._rev_eds, values)
  File "/home/simright/.pyenv/versions/pypi_fortranformat/lib/python3.7/site-packages/fortranformat/_output.py", line 116, in output
    sub_string = _compose_float_string(w, e, d, state, val, 'E')
  File "/home/simright/.pyenv/versions/pypi_fortranformat/lib/python3.7/site-packages/fortranformat/_output.py", line 253, in _compose_float_string
    sign_bit = val < 0
TypeError: '<' not supported between instances of 'NoneType' and 'int'

[hyperkang]

I have tried some local fixes. Will submit a PR later.

[hyperkang]

@brendanarnold I have submitted two PRs for writting int and float strings (with None value) respectively.

It would be good to add tests to cover the expected behaviors. But I need more time to understand how to add test.

[brendanarnold]

By convention, None represents a FORTRAN variable that hasn't been initialised. IIRC FORTRAN has default values for variables e.g. 0 for integers. It might be better instead to see what uninitialised variables print out and do that instead.

Running the following in GFortran ...

      PROGRAM FEDOUT

      REAL :: myreal
      LOGICAL :: mybool
      INTEGER :: myint

      WRITE(*, '(A, F10.1, A)') '[', myreal, ']'
      WRITE(*, '(A, L10, A)') '[', mybool, ']'
      WRITE(*, '(A, I10, A)') '[', myint, ']'

      STOP
      END

Gives the following output

[       0.0]
[         T]
[         0]

... so it looks like numerical values are initialised to '0' and logical to .TRUE.. I'll write some tests to cover this and update the code.

[hyperkang]

OK. So it looks we need to handle tailing None values in the list during "writing".

[brendanarnold]

Yep, here is a commit with tests showing how None values are handled de6d886

[hyperkang]

Yep, here is a commit with tests showing how None values are handled de6d886

Great. Please feel free to close my PRs.

[brendanarnold]

Now released as v1.2.0