[BEAM-1251] Change unicode --> six.text_type for Python 3

[cclauss]

unicode was removed in Python 3 because all str are Unicode so this PR:

• Changes all calls to unicode() into calls to six.text_type()
• Change remaining instances of unicode to six.text_type

[pabloem]

Python Precommits are broken for now (fix incoming). As advice, it'd be good for you to add a reviewer for your PR. I'll take a look in the next few days.

[cclauss]

How do I "add a reviewer"?

[pabloem]

There was some breakage in the Python SDK earlier. Could you rebase to master branch?

[pabloem]

Jenkins retest this please

[pabloem]

Nit: Lint is complaining about these:

 import apache_beam as beam
 from apache_beam.options.pipeline_options import PipelineOptions
-from apache_beam.transforms.display import (DisplayData, DisplayDataItem,
-                                            HasDisplayData)
+from apache_beam.transforms.display import DisplayData
+from apache_beam.transforms.display import DisplayDataItem
+from apache_beam.transforms.display import HasDisplayData

[pabloem]

Nit: Lint is complaining of the ordering of this import:

+import hamcrest as hc
 import six
-
-import hamcrest as hc
 from hamcrest.core.base_matcher import BaseMatcher

[cclauss]

This one complains when the hamcrest imports are split but also complains when they are put together. The desired order is also different than iSort. A bit difficult to find the correct order.

[cclauss]

So now it is saying:

 import hamcrest as hc
+import six
 from hamcrest.core.base_matcher import BaseMatcher
-
-import six

But when I move import six up in between the two hamcrest imports, it will complain that they can not be split. Trying that now...

[cclauss]

Now...

Running pylint for module apache_beam:
************* Module apache_beam.transforms.display_test
C: 27, 0: Imports from package hamcrest are not grouped (ungrouped-imports)

[cclauss]

Hopefully the # pylint: disable=ungrouped-imports and # pylint: enable=ungrouped-imports trick that I learned from @luke-zhu should placate pylint and this PR should be good to go.

[luke-zhu]

Glad that worked. I've added https://issues.apache.org/jira/browse/BEAM-3745 to track this.

[cclauss]

Can someone please advise how I can fix the failing test?

[luke-zhu]

Looks like the problem is unrelated. It's on the java branch and I saw that a different build on the same computer (beam 5) failed similarly here: https://builds.apache.org/job/beam_PreCommit_Java_GradleBuild/2529/console

You could also type in "retest this please" to the the checks again

[cclauss]

@aaltay Your review please?

[aaltay]

@cclauss LGTM. I will merge. As a side question, is it possible for us to update linter rules to avoid regressing?

How do I "add a reviewer"?
You could either tag someone as R: @altay, or add them as a reviewer in the UI on the top right.

[cclauss]

On update linter rules to keep from backsliding see #4798

On the reviewer tips, they only work if you have the commit bit which I do not have and do not need.

[aaltay]

On update linter rules to keep from backsliding see #4798

Thank you, will review it there.

On the reviewer tips, they only work if you have the commit bit which I do not have and do not need

You can add R; .. as a comment to the PR. Beam committer typically look for that tagging. You do not need to be a committer to do that.