findscu override keys

[roydobbins]

From the "findscu" manual:

[quote]
It is not possible to replace or insert attributes within sequences using the -k option.

[/quote]

Would anyone need this capability?

If anyone would like to use this feature, I am considering implementing it for an application I am working on.

Is there a good reason not to allow sequence insertion?

[Marco Eichelberg]

It is not possible to replace or insert attributes within sequences using the -k option. Would anyone need this capability?

There are a few cases where someone might want to specify an attribute within a sequence as part of a query. In particular in the case of a Modality Worklist query there are a few important query keys that are located in a sequence. Such queries currently can only be passed as a query file, but not constructed on the command line, i.e. the "-k" or "--key" option is not equally powerful as the query file facility.

Is there a good reason not to allow sequence insertion?

The only reason is that the application logic would be a bit more tricky and you would have to invent some convincing syntax for this. Say, somebody would like to add a certain sequence containing two items, with another sequence in the second (but not the first) item and an attribute with a specific value within an item of that nested sequence. How do you specify this on the command line? I would suggest something like

Code: Select all

--key OuterSequence[1].InnerSequence[0].Attribute=value

but the code evaluating this command line argument would not exactly be trivial.

[roydobbins]

Thanks for the comments.

Yes, I can see that this would not be trivial.

However, building a query file is also not trivial.

It could be done by having enough templates around to cover the commonly encountered queries and then programatically choosing the appropriate template and filling in the appropriate key values.

This approach is ok, but a little kludgy, so I was hoping for a more elegant solution. I suppose another approach would be a smart query builder, which takes the supplied keys and creates the complete query automatically. That would probably be the best generic solution, but I will have to put that off until a definite need arises for this sort of thing

I think you have convinced me that there are syntactic issues beyond the simple -k 0010,0010="abc" form for the command line override keys, so I will go with the template approach for now.

Thanks for the advice