faq.rst

F.A.Q.
======


qsub:  file must be an ascii script
-----------------------------------

This is mostly a copy and pasting error, especially when working with Microsoft
Word and other "editors".
Qsub jobs need to be in ascii format and cannot handle UTF-8 special characters.
To test if your jobfile/script contains non-ascii characters you can use the ``file`` command.
Converting can be done with your editor of choice or with the ``iconv`` program.

Example:

.. code-block:: bash

   # UTF-8, will probably produce a faulty job file
   root@master:$ file MRtrix3_TARDIS_0.3.sh
   MRtrix3_TARDIS_0.3.sh: Bourne-Again shell script, UTF-8 Unicode text executable

   # ASCII, will probably work
   root@master:$ file MRtrix3_preproc.sh
   MRtrix3_preproc.sh: Bourne-A gain shell script, ASCII text executable

   # convert from utf-8 to ascii will never work, but tell you the position of
   # the non-ascii character
   root@master:$ iconv -f utf-8 -t ascii MRtrix3_TARDIS_0.3.sh
   [...]
   #PBS -j oe
   #PBS iconv: illegal input sequence at position 3811


As you can see in the example above the character after ``#PBS`` following the
line ``#PBS -j oe`` could not be parsed to UTF-8. In this case the hyphen
character ``-`` was a UTF-8 special character instead. Just delete it and
replace the character with an actual hyphen and your script should
automatically become an ascii file.


heudiconv: UnicodeDecodeError
-----------------------------

When running Heudiconv, macOS users sometimes notice an error similar to this:

.. code-block:: python

      File "/src/heudiconv/heudiconv/bids.py", line 77, in populate_bids_templates
        populate_aggregated_jsons(path)
      File "/src/heudiconv/heudiconv/bids.py", line 109, in populate_aggregated_jsons
        json_ = load_json(fpath)
      File "/src/heudiconv/heudiconv/utils.py", line 182, in load_json
        data = json.load(fp)
      File "/opt/miniconda-latest/lib/python3.6/json/__init__.py", line 296, in load
        return loads(fp.read(),
      File "/opt/miniconda-latest/lib/python3.6/codecs.py", line 321, in decode
        (result, consumed) = self._buffer_decode(data, self.errors, final)
    UnicodeDecodeError: 'utf-8' codec can't decode byte 0xb0 in position 37: invalid start byte

Usually this indicates that Heudiconv tried to load json files that are in fact
`._` binary files created by `Apple's Finder
<https://apple.stackexchange.com/a/14981>`_. It seems that there is no way to
avoid their creation and you will have to clean them manually. Other
workarounds are setting your BIDS folder read-only before loading them in Finder or
avoiding Finder altogether.

To clean them try running **either**: ``dot_clean -m bids_folder`` on your Mac or
``find bids_folder -name '._*' -delete`` on a Linux/macOS.