Support loading unicode contained plugin in Python 3

[davidlatwe]

What's changed

Some of our plugins contain Unicode characters in plugin's label attribute or doc string. Those plugins which registered in plugin path can be discovered via Pyblish in Python 2 but fails in Python 3 due to bytes encoding issue.

The solution was simple, just change the open mode into "rb" in api.discover.

Example

1. Save a plugin with Unicode character

# /path/to/plugin/unicode_plugin.py

import pyblish.api

class UnicodePlugin(pyblish.api.InstancePlugin):

    order = pyblish.api.CollectorOrder
    label = "🤨"  # Unicode emoji

    def process(self, instance):
        return True

2. In Python 3, register the dir path and run pyblish.api.discover to see the test result

import pyblish.api

# Remove other plugins for testing
pyblish.api.deregister_all_plugins()
pyblish.api.deregister_all_paths()

# Register our only Unicode contained plugin
pyblish.api.register_plugin_path("/path/to/plugin")

# AssertionError will raised without this PR
assert len(pyblish.api.discover()) == 1

[BigRoy]

@davidlatwe could it be worth it potentially adding a test for this too or would it be overly verbose? Other than that, nice and clean solution!

[davidlatwe]

Yeah, I was thinking about adding a test case for this, too ! But maybe tomorrow, I am just about to go out now. ☺️

[mottosso]

Nice one. :) Yes, test would be good, just an additional file with lots of unicode should suffice. And if all existing tests pass already then that's already a giant confirmation it works.

[davidlatwe]

Test case added !

[mottosso]

Looks good, nice work. :)