Improve inventory target name resolution

[seanmil]

Update the target string name resolution behavior with two significant improvements:

• Do not perform wildcard lookups unless a wildcard character is present in the string, instead only performing exact-name matches.
• When performing a wildcard lookup, use fnmatch for improved speed and functionality.

[seanmil]

The documentation for the plan functions get_target and get_targets do not mention anything about wildcard support, but the Bolt documentation at https://www.puppet.com/docs/bolt/latest/running_bolt_commands#specify-targets-using-glob-matching mentions at the bottom that "Extended glob matching is only supported when specifying targets from the command-line interface. Plans do not support extended glob matching." - which seems to imply that basic glob matching should be supported.

[donoghuc]

I'll have to debug the tests (i'll do this in a separate PR). Assuming all the tests pass I think this is a great improvement. i agree with your assessment and interpretation of the intent for matching. Thanks!

[seanmil]

I'll have to debug the tests (i'll do this in a separate PR). Assuming all the tests pass I think this is a great improvement. i agree with your assessment and interpretation of the intent for matching. Thanks!

I made one more improvement to do an exact-match lookup instead of the slower wildcard if there are no wildcard characters.

In my ~3,300 target string name get_targets call it went from around 30s to 7s with the regex-to-fnmatch change and then from 7s to 0.035s with the exact-match logic change.

[donoghuc]

closing and re-opening to trigger tests.

[donoghuc]

Now that I got tests unblocked, you can see a couple cases we will need to address with this work.

[donoghuc]

Failing tests:

Failures:

  1) Bolt::Target when parsing userinfo strips brackets from ipv6 addresses in a uri
     Failure/Error: raise(WildcardError, target) if targets.empty?

     Bolt::Inventory::Inventory::WildcardError:
       Found 0 targets matching wildcard pattern ssh://[::1]:22
     # ./lib/bolt/inventory/inventory.rb:160:in `resolve_name'
     # ./lib/bolt/inventory/inventory.rb:180:in `block in expand_targets'
     # ./lib/bolt/inventory/inventory.rb:179:in `map'
     # ./lib/bolt/inventory/inventory.rb:179:in `expand_targets'
     # ./lib/bolt/inventory/inventory.rb:99:in `get_targets'
     # ./lib/bolt/inventory/inventory.rb:108:in `get_target'
     # ./spec/unit/target_spec.rb:177:in `block (3 levels) in <top (required)>'

  2) Bolt::Target when parsing userinfo returns set object with feature_set
     Failure/Error: raise(WildcardError, target) if targets.empty?

     Bolt::Inventory::Inventory::WildcardError:
       Found 0 targets matching wildcard pattern ssh://[::1]:22
     # ./lib/bolt/inventory/inventory.rb:160:in `resolve_name'
     # ./lib/bolt/inventory/inventory.rb:180:in `block in expand_targets'
     # ./lib/bolt/inventory/inventory.rb:179:in `map'
     # ./lib/bolt/inventory/inventory.rb:179:in `expand_targets'
     # ./lib/bolt/inventory/inventory.rb:99:in `get_targets'
     # ./lib/bolt/inventory/inventory.rb:108:in `get_target'
     # ./spec/unit/target_spec.rb:192:in `block (3 levels) in <top (required)>'

Finished in 18.04 seconds (files took 1.62 seconds to load)
[153](https://github.com/puppetlabs/bolt/actions/runs/8713085630/job/23900793508?pr=3291#step:7:154)6 examples, 2 failures

Failed examples:

rspec ./spec/unit/target_spec.rb:176 # Bolt::Target when parsing userinfo strips brackets from ipv6 addresses in a uri
rspec ./spec/unit/target_spec.rb:191 # Bolt::Target when parsing userinfo returns set object with feature_set

[seanmil]

Failed examples:

rspec ./spec/unit/target_spec.rb:176 # Bolt::Target when parsing userinfo strips brackets from ipv6 addresses in a uri
rspec ./spec/unit/target_spec.rb:191 # Bolt::Target when parsing userinfo returns set object with feature_set

@donoghuc I've fixed these with what I believe to be a correct fix (if it looks like a URI, don't try wildcard matching on it) but one seemingly unrelated test failed. Maybe it would work if the CI run was re-started.

[donoghuc]

triggering a CI run by opening and then closing this PR.

[donoghuc]

Ah, so it looks like this actually exposed a small issue in our tests #3318

I think the fact that this now results in raising a WildcardError instead of letting #{target} through is an improvement. Thanks!

[mcdonaldseanp]

@seanmil I just pushed the test fix from Cas to main, if you rebase this on that and we see a green run we'll merge.

[seanmil]

@seanmil I just pushed the test fix from Cas to main, if you rebase this on that and we see a green run we'll merge.

All green (finally!), thanks!

[donoghuc]

Thank you!