PR: Add options to include user's local PYTHONPATH and environment variables

[mrclary]

Description of Changes

• Wrote at least one-line docstrings (for any new functions)
• Included a screenshot or animation (if affecting the UI, see Licecap)

Added preference to the Python Interpreter section allowing all platforms and launch methods to select whether a user's system PYTHONPATH and environment variables are used for Consoles.

Spyder's PYTHONPATH Manager list behavior is unchanged.

Fixes #14843

New Behavior

• If "Use system PYTHONPATH" is deselected: paths in PYTHONPATH Manager are appended to the Console's sys.path
• If "Use system PYTHONPATH" is selected: in addition to the above, the paths in the user's local PYTHONPATH environment variable are appended to the Console's sys.path. On macOS and Linux, the user's local PYTHONPATH is determined by /etc/profile, ~/.bash_profile, ~/.bash_login, and ~/.profile. On Windows, the user's local PYTHONPATH is extracted from the registry query reg query HKEY_CURRENT_USER\Environment.
• If "Use system environment variables" is deselected: only the following variables are passed to the Console's os.environ.
  • PATH and HOME on macOS
  • PATH, HOME, DISPLAY and XAUTHORITY on Linux
  • 'PATH', 'SYSTEMROOT', 'SYSTEMDRIVE', and 'USERPROFILE' on Windows
• If "Use system environment variables" is selected, all of the user's local environment variables, except PYTHONPATH, are passed to the Console's os.environ. On macOS and Linux, these are determined by /etc/profile, ~/.bash_profile, ~/.bash_login, and ~/.profile. On Windows these are determined by the registry query reg query HKEY_CURRENT_USER\Environment

Affirmation

By submitting this Pull Request or typing my (user)name below,
I affirm the Developer Certificate of Origin
with respect to all commits and content included in this PR,
and understand I am releasing the same under Spyder's MIT (Expat) license.

I certify the above statement is true and correct:
@mrclary

[mrclary]

This PR and #14853 are mutually exclusive.

[pep8speaks]

Hello @mrclary! Thanks for updating this PR. We checked the lines you've touched for PEP 8 issues, and found:

There are currently no PEP 8 issues detected in this Pull Request. Cheers! 🍻

Comment last updated at 2021-06-19 19:51:18 UTC

[ccordoba12]

Thanks @mrclary for your work on this! I left an initial review for you.

[mrclary]

I can't for the life of me figure out how to fix this problem...
I've gotten all tests to pass except for slow tests on Linux.
I've got the environment setup on my Ubuntu VM as close to identical to the Github test environment as possible, compared all the conda package versions, and cannot reproduce the error locally. The tests pass locally.

Failed test_run_cython_code

=================================== FAILURES ===================================
_____________________________ test_run_cython_code _____________________________

main_window = <spyder.app.mainwindow.MainWindow object at 0x7f5ddcb26f50>
qtbot = <pytestqt.qtbot.QtBot object at 0x7f5ddc049c50>

    @pytest.mark.slow
    @flaky(max_runs=3)
    @pytest.mark.skipif(
        (os.name == 'nt' or sys.platform == 'darwin' or
         parse_version(ipy_release.version) == parse_version('7.11.0')),
        reason="Hard to test on Windows and macOS and fails for IPython 7.11.0")
    def test_run_cython_code(main_window, qtbot):
        """Test all the different ways we have to run Cython code"""
        # ---- Setup ----
        # Get a reference to the code editor widget
        code_editor = main_window.editor.get_focus_widget()
    
        # ---- Run pyx file ----
        # Load test file
        main_window.editor.load(osp.join(LOCATION, 'pyx_script.pyx'))
    
        # Run file
        qtbot.keyClick(code_editor, Qt.Key_F5)
    
        # Get a reference to the namespace browser widget
        nsb = main_window.variableexplorer.current_widget()
    
        # Wait until an object appears
        qtbot.waitUntil(lambda: nsb.editor.source_model.rowCount() == 1,
>                       timeout=COMPILE_AND_EVAL_TIMEOUT)
E       AssertionError: waitUntil timed out in 30000 miliseconds
E       assert not True
E        +  where True = <function QtBot.waitUntil.<locals>.timed_out at 0x7f5dd3f1e4d0>()

/home/runner/work/spyder/spyder/spyder/app/tests/test_mainwindow.py:1104: AssertionError
----------------------------- Captured Qt messages -----------------------------
    QtWarningMsg: Scenegraph already initialized, setBackend() request ignored
--------------------------- Captured stdout teardown ---------------------------

In [1]: runfile('/home/runner/work/spyder/spyder/spyder/app/tests/pyx_script.pyx', wdir='/home/runner/work/spyder/spyder/spyder/app/tests')
gcc: fatal error: cannot execute ‘cc1’: execvp: No such file or directory
compilation terminated.

In [2]: 
---------------------------- Captured stderr setup -----------------------------
<class 'spyder.app.tests.spyder_boilerplate.spyder.plugin.SpyderBoilerplate'>: A plugin with section "spyder_boilerplate" already exists!
Traceback (most recent call last):
  File "/home/runner/work/spyder/spyder/spyder/app/mainwindow.py", line 917, in setup
    self.register_plugin(plugin_instance, external=True)
  File "/home/runner/work/spyder/spyder/spyder/app/mainwindow.py", line 237, in register_plugin
    plugin._register()
  File "/home/runner/work/spyder/spyder/spyder/api/plugins.py", line 884, in _register
    self._conf.register_plugin(self)
  File "/home/runner/work/spyder/spyder/spyder/config/manager.py", line 119, in register_plugin
    'exists!'.format(conf_section))
RuntimeError: A plugin with section "spyder_boilerplate" already exists!

test_move_to_first_breakpoint for pip slow linux has the same stderr setup error: RuntimeError: A plugin with section "spyder_boilerplate" already exists!

I just can't get these tests to fail locally, so I can't figure out what is going on.
@spyder-ide/core-developers , can you provide any insight or recommendations?

[dalthviz]

@mrclary I think this is related with the fake plugin I added and the reuse of the mainwindow instance in the tests. I will check 👍 (probably we need to add some sort of condition/validation so the addition of the fake plugin entrypoint is only done once)

[mrclary]

@dalthviz, thanks for looking into the fake plugin, perhaps that is the issue. However, I am curious why that doesn't manifest in the mac and windows tests also, or on my local linux tests.

[mrclary]

From conversation in #16200, remember to consider how sys.path is applied when using "Execute in an external systerm terminal" run option.

[mrclary]

Closing in favor of #16429