py3 issues

[Will Moore]

Batch_ROI_Export failing because of unicode um characters at

 File "./script", line 335, in run_script
    result = batch_roi_export(conn, script_params)
  File "./script", line 280, in batch_roi_export
    file_ann = write_csv(conn, export_data, script_params, symbol)
  File "./script", line 238, in write_csv
    csv_file.write("\n".join(csv_rows))
UnicodeEncodeError: 'ascii' codec can't encode character '\xb5' in position 64: ordinal not in range(128)

Populate_Metadata.py - opening script dialog gives:

omero.ValidationException: exception ::omero::ValidationException
{
    serverStackTrace = 
    serverExceptionClass = 
    message = Can't find params for 266058!
No stdout.
stderr is in file 266442:
---------------------------------
Traceback (most recent call last):
  File "./script", line 43, in <module>
    from omero_metadata.populate import ParsingContext
  File "/opt/omero/server/venv3/lib64/python3.6/site-packages/omero_metadata/populate.py", line 87
    Report bugs to [e-mail]""" % (error, cmd, cmd)
                                                          ^
SyntaxError: invalid syntax

[Sébastien Besson]

From the former issue it looks like we need to specify utf-8 encoding wherever appropriate similarly to ome/omero-py#110. Looking at https://python-future.org/compatible_idioms.html#file-io-with-open, it might be possible to have a unified Py2/Py3 idiom:

from io import open
...
    with open(file_name, 'w', encoding='utf-8') as csv_file:
...

The latter issue is related to the fact the omero-metadata plugins needs to be upgraded to Python 3 - see in-progress work in ome/omero-metadata#36

[Will Moore]

There was a workaround in d9119de to only write encoded bytes to csv, because writing unicode to csv fails on the py3-ci server due to lack of UTF-8 encoding as described ome/omero-figure#343 (comment) and ome/omero-figure#351 (comment)