[FEAT] Add something to let know something is happening during analysis

[thomasboni]

Is your feature request related to a problem? Please describe.

When I launch plumber analyze, a few lines are displayed, then nothing happens during ~15 seconds. I don't know if it's working in the background or if there is a bug.

➜  backend2 git:(main) plumber analyze
Auto-detected GitLab URL: https://gitlab.com
Auto-detected project: r2devops/jobs

  ██████╗ ██╗     ██╗   ██╗ ███╗   ███╗██████╗ ███████╗██████╗ 
  ██╔══██╗██║     ██║   ██║ ████╗ ████║██╔══██╗██╔════╝██╔══██╗
  ██████╔╝██║     ██║   ██║ ██╔████╔██║██████╔╝█████╗  ██████╔╝
  ██╔═══╝ ██║     ██║   ██║ ██║╚██╔╝██║██╔══██╗██╔══╝  ██╔══██╗
  ██║     ███████╗╚██████╔╝ ██║ ╚═╝ ██║██████╔╝███████╗██║  ██║
  ╚═╝     ╚══════╝ ╚═════╝  ╚═╝     ╚═╝╚═════╝ ╚══════╝╚═╝  ╚═╝
  CI/CD Compliance Scanner for GitLab Pipelines
  Join our community: https://getplumber.io/discord

Using configuration: .plumber.yaml
Analyzing project: r2devops/jobs on https://gitlab.com
Using local CI configuration (specify --branch to force upstream CI config fetch): /Users/thomasboni/workspace/gitlab.com/r2devops/backend2/.gitlab-ci.yml

Describe the solution you'd like

Add something to let me know that something is working in background. I don't what but we can find good examples with other cli tools

[slyt3]

Reproduced with timing logs, delay is in Fetching GitLab CI component catalog... (GitLab GraphQL), not in general analyze flow.

In my run:

• project fetch: ~0.82s
• CI merge fetch: ~0.84s
• component catalog fetch: ~36.91s (3 retries + timeout)
• image + controls: ~3s total

So the skip-when-no-component-includes optimization helps a lot for repos without component includes, but repos with component includes can still be slow when GitLab GraphQL is timing out.

[Joseph94m]

It's the datacollections not the controls that take a lot of time.

Given the current implementation, datacollections run regardless of which controls are disabled. There is room for improvement there but that should be another issue.

I'll think of an implementation proposition and create the issue next week.
Basically controls will have to declare which data they need.

At the start of an analysis, the CLI knows the set of controls that must run, deduces the set of data need, builds the set of datacollections necessary to obtain that data, finally only runs the datacollections needed.