Closed musicEnfanthen closed 6 years ago
Interesting... it's possibly the regex I use to identify the token boundary. Can you share the ontology you're using so I can try and reproduce locally?
I just tried to reproduce by adding a predicate E1_CRM_Entity to the foaf_override.json test file (in /spec) and it works fine. I also checked the regex and that won't be the cause. Which version of the completer are you using? I saw this behavior from time to time in early versions but haven't seen it recently.
Ok, I see. Sorry, I did not recognize that the suggestions in question did not come from the turtle-completer but from Atoms in-built autocomplete-plus package.
Found a similar issue on another repository (https://github.com/autocomplete-python/autocomplete-python/issues/364) that supposes that this issue has its very grounds in atoms autocomplete-plus package https://github.com/atom/autocomplete-plus/issues/950.
So sorry for the false alert.
Thanks for digging into this! You prompted me to add some more unit tests around the edge cases so still valuable :-)
When the value to be completed starts with a combination of letters and numbers (like in #CIDOCs
E1_CRM_Entitiy
;F22_Self-Contained_Expression
), the autocompleted result will double the letter/number combination like so:E1E1_CRM_Entity
,F22F22_Self-Contained Expression
.The underscore is irrelevant for this issue,
E1CRM
results inE1E1CRM
, too.