Download actions job logs from API

[lunny]

Related to #33709, #31416

It's similar with https://docs.github.com/en/rest/actions/workflow-jobs?apiVersion=2022-11-28#download-job-logs-for-a-workflow-run--code-samples.

This use job_id as path parameter which is consistent with Github's APIs.

[wxiaoguang]

I am sure you just copied & pasted the code from somewhere without understanding what each line does.

[ChristopherHX]

Works for me.

While testing I remembered workflow_job webhook uses jobid instead of job index for api url, to have the same field to url mapping for the url field as GitHub Rest.

e.g. http://localhost:3000/api/v1/repos/test/private-repo/actions/runs/842/jobs/373 (prefix of this logs endpoint)

instead of http://localhost:3000/api/v1/repos/test/private-repo/actions/runs/842/jobs/0.

And logs is below this path.

The swagger documention tells me I can use the id that is 404.

This use run_id as path parameter which is consistent with other APIs but not web routers. Maybe all web routers should change from run_index to run_id to keep consistent.

I would prefer run_id and job_id in both web routers and api. However I know that this is going to break existing commit status reports

I guess workflow_job webhook is currently a good way to obtain all ids to construct this url, we might need a workflow_run/workflow_job api route to make this really usable from scripts.

EDIT

Sorry some parts are incorrect from my side...

The gh compatible api path of workflow_job is the following, in this case only a id makes sense.

http://localhost:3000/api/v1/repos/test/private-repo/actions/jobs/373

[ChristopherHX]

We were looking at the wrong section of the gh docu.

You are implementing this endpoint https://docs.github.com/en/rest/actions/workflow-jobs?apiVersion=2022-11-28#download-job-logs-for-a-workflow-run--code-samples under a different url.

The one from GitHub has no runs part in the url and uses the jobid instead of index

[lunny]

job_id is a string which is from the workflow not the id stored in the database.
ref https://docs.github.com/en/enterprise-cloud@latest/actions/writing-workflows/workflow-syntax-for-github-actions#jobsjob_id

[ChristopherHX]

job_id is a string which is from the workflow not the id stored in the database.

GitHub is good in mixing names, yes job_id in workflow yaml is a string.

job_id in workflow_job api object is a number from the database see here:

ref https://docs.github.com/en/rest/actions/workflow-jobs?apiVersion=2022-11-28#get-a-job-for-a-workflow-run

The example has a really big number, that is a global database id I guess

"id": 399444496,

GitHub Actions users have hard time to correlate a workflow execution from the runner to a workflow_job api id

[ChristopherHX]

Looks good to me, left some non functional suggestions