For a number of reasons I won’t go into here, LLM suggestions for TiddlyWiki coding are consistently bad.
In this particular case, the suggested $transclude widget usage:
<$transclude tiddler="..." index="..."/>
can only access top-level "key":"value" JSON data, like this:
{
"a": "one",
"b": "",
"c": "three"
}
If a JSON tiddler contains nested objects (e.g., “d” in your example JSON), those deeper level JSON data values cannot be accessed by the $transclude widget. Note that if a JSON tiddler actually contained a top-level index named “d.e”, the $transclude syntax would be able to retrieve it, since “d.e” is just an index key that happens to contain a “.” as part of the key name.
Still… it might be a nice TWCore enhancement if the syntax suggested by Gemini actually could produce results for nested JSON objects. What would be needed is a “fallback” approach:
- The
$transcludewidget processing would first attempt to find a top-level key with the specified “d.e” key name (as it currently does) - Then, if that doesn’t yield a result, the TWCore could split the “d.e” key name on the “.” and use each part to process the tiddler’s text content as if a
[[tiddlername]get[text]jsonget[d],[e]]filter were being applied.
This might even be relatively easy to implement by simply calling on the existing jsonget filter code to perform the extended JSON processing. I have no idea if this approach would have any significant performance issues, but I think it’s worth considering.
@jeremyruston … what do you think?
-e