fix: human time must not hard set the timezone

[sdenef-adeo]

Hello,

utils.get_human_time() function format a UNIX timestamp as string to be stored into BigQuery.
But it replaces the timezone without dealing with the UTC time.

Using datetime.fromtimestamp() do the job well.

Tests have been updated accordingly.

Regards,
Sébastien

[sdenef-adeo]

Hello @lvaylet,

I don't understand why some checks failed as it works well and I'm pretty sure the method signature is correct.
Any help will be really appreciated.

Regards,
Sébastien

[lvaylet]

Hello @sdenef-adeo, I will take a look later today.

[lvaylet]

@sdenef-adeo The pytype stage of make lint completes successfully on my machine after removing timestamp=:

    dt_tz = datetime.fromtimestamp(timestamp, tz=to_zone)

But then the safety stage fails because of CVEs in GitPython and certifi.

[sdenef-adeo]

Hello @lvaylet,

👌
AFAIU those CVE are not due to the code provided here. Is there any other PR I need to merge into this one to fix those CVE?

[lvaylet]

I quickly checked yesterday but was unable to spot the exact source(s). These CVEs were probably reported recently. I wil work on a separate PR tomorrow and let you know when you can merge.

[lvaylet]

@sdenef-adeo I just created #352 and will start investigate soon.

[lvaylet]

@sdenef-adeo Done. Can you merge fb35d87 into your PR?

[sdenef-adeo]

merged

[lvaylet]

Then I guess you also have to remove timestamp= from your code, as highlighted above.

[lvaylet]

Thanks @sdenef-adeo. Now that all checks complete successfully, let me reproduce the (bad) behavior and check this PR fixes it.

[lvaylet]

I just confirmed that the current code does not return the expected values.

For example:

$ python
Python 3.9.15 (main, Nov  5 2022, 10:20:53) 
[GCC 10.2.1 20210110] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> from slo_generator.utils import get_human_time
>>> get_human_time(1702660513.987654, timezone='Europe/Paris')
'2023-12-15T17:15:13.987654+01:00'
>>> get_human_time(1565092435, timezone='Europe/Paris')
'2019-08-06T11:53:55.000000+02:00'

According to Epoch & Unix Timestamp Conversion Tools, the first call should return '2023-12-15T18:15:13.987654+01:00'. The second call should return '2019-08-06T13:53:55.000000+02:00'.

Your version returns the right values.

Can you rebase your branch as it is now out-of-date? Then I will merge this PR.

[sdenef-adeo]

for sure @lvaylet

It should be OK now.

[lvaylet]

Not quite. GitHub still reports 'This branch is out-of-date with the base branch'. I will move forward anyway, as these changes do not overlap with anything in recent PRs.

[lvaylet]

Unfortunately, GitHub is preventing me from moving forward.

Your master branch seems to lag behind by 2 commits:

Can you update your fork, then rebase your adeo:fix/timestamp_human_tz branch too?

[sdenef-adeo]

I understand. It should be due to my "master branch" on the fork repo is out-of-date. I will do.

[lvaylet]

All good. Thanks for your patience @sdenef-adeo!