[AID] Handling of semantic annotations #54
Conversation
|
|
||
| if (Array.isArray(tdContext)) { | ||
| if (tdContext.length > 3) { | ||
| // if it is longer than 3, assume 3rd item is the object. This offset is needed as node-wot always adds @language at the end |
There was a problem hiding this comment.
This offset can be avoided if we do not normalize the TD when parsing. In any case, I find it a bit weird that node-wot changes the context.
Also this really assumes that we have the TD contexts (v1 or v1.1) and then an object with prefixes.
There was a problem hiding this comment.
Mhh, not sure what is meant by "node-wot changes the context". Should we open an issue on node-wot side?
Anyhow, it should work in any order ... but I do recall that in the TD spec there is a order defined... maybe that is the reason?
There was a problem hiding this comment.
In general I think this part of the code assumes some characteristics that might change over time in node-wot. Not sure if this future proof...
There was a problem hiding this comment.
I also feel that this is not future-proof but I have thought of a small improvement: Going through all object items in the array and merging them into one object.
node-wot is not doing anything against the TD spec but it always adds an object with @language in it. So the TD I provide is changed after the parsing.
| @@ -1213,12 +1213,15 @@ export class AssetInterfacesDescription { | |||
| for (const propertyKey in td.properties) { | |||
| const property: PropertyElement = td.properties[propertyKey]; | |||
|
|
|||
| // we get all the property keys and remove known keys by the TD spec at each if clause | |||
| let propertyKeys: any = Object.keys(property); | |||
There was a problem hiding this comment.
This propertyKeys handling is the part I find a big ugly but I could not find a better solution. I went initially for a full json-ld processing but it me messed up... So each time the conversion finds a known TD term, it gets removed from the array. In the end, from the unknown terms, the ones with with a : are taken for adding their semantic id with some basic json-ld processing
|
|
||
| if (Array.isArray(tdContext)) { | ||
| if (tdContext.length > 3) { | ||
| // if it is longer than 3, assume 3rd item is the object. This offset is needed as node-wot always adds @language at the end |
There was a problem hiding this comment.
In general I think this part of the code assumes some characteristics that might change over time in node-wot. Not sure if this future proof...
Co-authored-by: danielpeintner <[email protected]>
If a TD has a semantic annotation, it is simply ignored in the conversion. This PR adds a basic support in this direction. I am not exactly satisfied since there can be many corner cases. The worst is if there is an additional context without a prefix. That would mean looking into the context for each possible unknown key. I have added a good amount of comments in the code but also adding now some in the PR review.