Add project support in jobs

[YevheniiSemendiak]

closes #2948

Some assumptions made:

• while constructing JobDescription from the API response, if no project_name is given, we use job owner
• in shell completion, we filter and suggest jobs based on project_name instead of owner
• we'll need to patch those URI-parsing functions in a separate PR https://github.com/neuro-inc/neuro-cli/compare/ys/job-project?expand=1#diff-d5563da4783e29f5c8dca3b4ffd9efb0915ac0af5f5dd820e1ad4f3501041845R58

[codecov]

Codecov Report

Patch coverage: 92.00% and project coverage change: +0.06 🎉

Comparison is base (9db3432) 73.63% compared to head (fe5b9b2) 73.69%.

❗ Current head fe5b9b2 differs from pull request most recent head a92e29f. Consider uploading reports for the commit a92e29f to get more accurate results

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #2955      +/-   ##
==========================================
+ Coverage   73.63%   73.69%   +0.06%     
==========================================
  Files          70       70              
  Lines       12500    12502       +2     
  Branches     2303     2302       -1     
==========================================
+ Hits         9204     9213       +9     
+ Misses       2975     2973       -2     
+ Partials      321      316       -5     

Impacted Files	Coverage Δ
neuro-cli/src/neuro_cli/parse_utils.py	89.50% <ø> (ø)
neuro-sdk/src/neuro_sdk/_url_utils.py	86.60% <ø> (ø)
neuro-cli/src/neuro_cli/job.py	43.37% <66.66%> (+0.15%)	⬆️
neuro-cli/src/neuro_cli/click_types.py	75.51% <100.00%> (-0.10%)	⬇️
neuro-cli/src/neuro_cli/formatters/jobs.py	85.09% <100.00%> (+1.01%)	⬆️
neuro-sdk/src/neuro_sdk/_jobs.py	81.25% <100.00%> (+0.43%)	⬆️

... and 3 files with indirect coverage changes

Help us with your feedback. Take ten seconds to tell us how you rate us. Have a feature suggestion? Share it here.

☔ View full report in Codecov by Sentry.
📢 Do you have feedback about the report comment? Let us know in this issue.

[dalazx]

I am not entirely sure about this. we created default projects for existing users just once. any new users will have to create their projects manually. thus the names of the projects may not be equal to user names (there's a chance of conflicts we discussed earlier). so self._config.username might be confusing as it may result in permission errors. additionally, we already handle this on the backend side. wdyt?

[YevheniiSemendiak]

Yes, API wouldn't allow users to create jobs not in their projects or w.o. projects at all, we are safe on the backend side.

As for SDK/CLI, we could either make the project_name parameter of Jobs.run and Jobs.start obligatory, OR rely on self._config.project_name and raise an exception at _job_to_api stating smth around "Project is not specified, make sure you have created one, or reload configuration before running the jobs". I'm more for 2nd approach, wdyt?

[andriihomiak]

From a user experience point of view, 2 is preferred. Most users rarely switch projects, and having to constantly retype even though it rarely changes would be a hassle.

[andriihomiak]

This is similar in a way to clusters - most users will never access more than one

[dalazx]

these all seem valid points. my only concern is us explicitly using self._config.username. by the time this code executes all existing users will have their default project created and even pulled from /config. otherwise asking to create a project seems fine.

[YevheniiSemendiak]

in 890658a I've added one extra propery of config named project_name_or_raise which raises exception

The current project is not selected. Please create one with 'neuro admin add-project', or switch to the existing one with 'neuro config switch-project'.

So this property to start a job if no project parameter is specified.
I'm not using project_name property directly, since when we add the user to the cluster, it won't have any project yet.
wdyt, @dalazx ?

[dalazx]

👍