opcua: Fixes issues with computing "information" based on path

[wilterdinkrobert]

Bugfix

Timeseries with nodeid paths containing \\ in the string part failed with re.error bad escape \... at position .... The fullmatch(path, config) would compile the path part as a regex, but cannot be assumed to be correctly escaped (hence the re.error). Upon closer inspection the rather complicated code can be removed completely by storing the information with the subscription when it is first made. Functionality from configuration pov should be unchanged/equivalent.

Test configuration:

        "timeseries": [
          {
            "key": "MachineRt_Error_Status",
            "path": "${ns=2;s=MachineRt\\Error\\Status}"
          },{
            "key": "MachineRt_Cycle_Status",
            "path": "${ns=2;s=MachineRt\\Cycle\\Status}"
          }]

Tested with several small OPCUA servers.

[imbeacon]

Hi @wilterdinkrobert,
Thank you for your interest and participating in the project life. I have checked the difference and could you confirm that configuration like the following one will still work correctly? Or the same path, but from the Root object on the server?

        "timeseries": [
          {
            "key": "MachineRt_Error_Status",
            "path": "${MachineRt\\.Error\\.Status}"
          },{
            "key": "MachineRt_Cycle_Status",
            "path": "${MachineRt\\.Cycle\\.Status}"
          }]

[wilterdinkrobert]

Yes, I did test the partial browse_path. I've now also tested with the full browse_path (from Root) and encountered some more bugs/inconsistencies (not related to the previous commit). Could you please confirm that the tests in 8f67463 match the intended behavior?

[imbeacon]

Yes, looks good.

[wilterdinkrobert]

@imbeacon Thanks for confirming. I just pushed the last cleanup commit. I think it's ready for merging now, what do you think?