[time] Cross-platform support for device timezones via tzlocal

[michaelfromyeg]

I tried to use the time server on Windows, and noticed a timezone error. Windows seems to report the timezone in a strange format...

# windows
>>> import datetime
>>> tzinfo = datetime.datetime.now().astimezone(tz=None).tzinfo
>>> tzinfo
datetime.timezone(datetime.timedelta(days=-1, seconds=61200), 'Pacific Daylight Time')

# linux
>>> import datetime
>>> tzinfo = datetime.datetime.now().astimezone(tz=None).tzinfo
>>> tzinfo
datetime.timezone(datetime.timedelta(days=-1, seconds=61200), 'PDT')

These kinds of values didn't play nice with ZoneInfo. This PR adds tzlocal to fix that.

One thing: I couldn't figure out how to test this reliably. Let me know if you'd like me to add tests.

cc @maledorak for viz!

edit: also #320 is probably worth another glance 👀

Description

Server Details

• Server: time
• Changes to: the get_local_tz helper

Motivation and Context

• Fixes a bug when starting the server on windows (zoneinfo._common.ZoneInfoNotFoundError: 'No time zone found with key Pacific Daylight Time') via tzlocal

How Has This Been Tested?

• Tested manually on Windows; working as expected

Breaking Changes

No

Types of changes

• Bug fix (non-breaking change which fixes an issue)
• New feature (non-breaking change which adds functionality)
• Breaking change (fix or feature that would cause existing functionality to change)
• Documentation update

Checklist

• I have read the MCP Protocol Documentation
• My changes follows MCP security best practices
• I have updated the server's README accordingly
• I have tested this with an LLM client
• My code follows the repository's style guidelines
• New and existing tests pass locally
• I have added appropriate error handling
• I have documented all environment variables and configuration options

Additional context

[maledorak]

Hey @michaelfromyeg , thanks for cc :)
Sadly, I don't have a Windows machine, and I can't test this code on Windows :/

[michaelfromyeg]

ah all good!

[olaservo]

Thanks for the PR! I'm wondering if we should try to combine this approach and the approach in #1272 to cover all our bases (I left a comment in that one).

[michaelfromyeg]

I trust your judgement on which is best! Not sure how much fallback-ing is necessary... but better safe than sorry?

Candidly, tzlocal was what came up first when I searched up the issue (via Claude!).

[olaservo]

@michaelfromyeg since your PR includes some doc formatting updates too and was created first, let's do this:

• Could you add some tests to the existing tests to cover a few cases related to this fix?
• Then we can merge this PR and the other PR can get updated to add any additional stuff not covered by this PR.

[olaservo]

Hi, was running the tests for this locally and ran into some errors related to McpError so I left a few suggestions to resolve those.

Also a suggestion for a test to add that would check for valid vs invalid overrides:

@pytest.mark.parametrize(
    "test_time,local_tz_override,expected_error",
    [
        # Test Windows format timezone override
        (
            "2024-01-01 12:00:00+00:00",
            "America/Los_Angeles",  # What Windows "Pacific Standard Time" should map to
            None,
        ),
        # Test direct IANA name
        (
            "2024-01-01 12:00:00+00:00", 
            "Europe/Berlin",  # What "Mitteleuropäische Zeit" should map to
            None,
        ),
        # Test invalid timezone
        (
            "2024-01-01 12:00:00+00:00",
            "Invalid/Timezone",
            "Invalid timezone: 'No time zone found with key Invalid/Timezone'",
        ),
    ],
)
def test_get_local_tz_override(test_time, local_tz_override, expected_error):
    """Test local timezone override functionality"""
    with freeze_time(test_time):
        if expected_error:
            with pytest.raises(McpError, match=expected_error):
                get_local_tz(local_tz_override)
        else:
            result = get_local_tz(local_tz_override)
            assert isinstance(result, ZoneInfo)
            assert str(result) == local_tz_override

[olaservo]

Suggestion to add try/catch if the local override results in an exception, plus it looks like McpError expects an object with a message and not a string:

Suggested change

	if local_tz_override:
	return ZoneInfo(local_tz_override)
	if local_tz_override:
	try:
	return ZoneInfo(local_tz_override)
	except Exception as e:
	error_data = ErrorData(
	code=INVALID_PARAMS,
	message=f"Invalid timezone: {str(e)}"
	)
	raise McpError(error_data)

[olaservo]

You will also need to add ErrorData and INVALID_PARAMS fom mcp.types, for example:

from mcp.types import Tool, TextContent, ImageContent, EmbeddedResource, ErrorData, INVALID_PARAMS

[olaservo]

Suggested change

	raise McpError("Could not determine local timezone - tzinfo is None")
	error_data = ErrorData(
	code=INVALID_PARAMS,
	message="Could not determine local timezone - tzinfo is None"
	)
	raise McpError(error_data)

[olaservo]

I can't leave comments or suggestions on unchanged lines, but I think you also want to change this in the get_zoneinfo function:

    except Exception as e:
        error_data = ErrorData(
            code=INVALID_PARAMS,
            message=f"Invalid timezone: {str(e)}"
        )
        raise McpError(error_data)