It's possible but not particularly straightforward in current versions of FME. Not sure whether the same sort of workflow would work in 2015

Ragged edge tables are seldom fun to work with, can you refactor the problem to (say) output a worksheet per ID value and keep Secondary ID in one column?

Potentially, but I may have to change things further down the line & we already have a script set up to work with such a ragged edge table. Perhaps we need to make some changes in that regards though.

Now I know I could do that with a ListBuilder/ListConcatenator, and it seems it'd be wise for me to use a ListElementCounter to get an idea of how many columns I would need.

Anyhow, I've built a very non-elegant solution. I'm just building a list and setting up an AttributeCreator to output a new attribute for each list element number.

e.g., AttributeName = Secondary_ID_1, Value = @Value(_list{0}.Secondary_ID)

AttributeName = Secondary_ID_2, Value = @Value(_list{1}.Secondary_ID)

AttributeName = Secondary_ID_3, Value = @Value(_list{2}.Secondary_ID)

...and so on until I've created enough new attributes to hold the max count from the ListElementCounter.

Perhaps a more elegant solution will come to mind & if so I'll post here.

Thanks all for the quick feedback.

This is how I would do it in 2018

and the python to avoid having to manually expose attributes

import fme
import fmeobjects

def create_schema(feature):
    attrlist = feature.getAttribute('_list{}.Sec_ID_Name')
    feature.setAttribute('attribute{'+str(0)+'}.name',"ID")
    feature.setAttribute('attribute{'+str(0)+'}.fme_data_type','fme_char(20)')
    for i, val in enumerate(attrlist):
        feature.setAttribute('attribute{'+str(i+1)+'}.name',val)
        feature.setAttribute('attribute{'+str(i+1)+'}.fme_data_type','fme_char(20)')

import fme
import fmeobjects

def create_schema(feature):
    attrlist = feature.getAttribute('_list{}.Sec_ID_Name')
    feature.setAttribute('attribute{'+str(0)+'}.name',"ID")
    feature.setAttribute('attribute{'+str(0)+'}.fme_data_type','fme_char(20)')
    for i, val in enumerate(attrlist):
        feature.setAttribute('attribute{'+str(i+1)+'}.name',val)
        feature.setAttribute('attribute{'+str(i+1)+'}.fme_data_type','fme_char(20)')

Thanks again.

@bwasserstein

Listbuilder on ID, explode. Increase value of elemtn index by 1.

Concatenate this with Secondary_ID_.

Create attribute @Value(_result) = @Value(Secondary ID)

Expose Secondary_ID_1 to Secondary_ID_n (create list in excel by increment and import in exposer)....

aaah..can't import in fme_2015. Well hope you don't have hundreds of ID nrs then. (upgrade FME?)

aggregator on ID, drop attributes, use attriubtus to concatenate on Secondary_ID_1 to Secondary_ID_n with no separator.

Now you're done

here is result

If you populate Secondary IDs grouping by primary ID into a list, you can just rename each element to a desired name with the AttributeRenamer.

@takashi

Of course that's possible.

Just not very attractive because then you can't import a attribute names . Unless you create a mapping first and import that, which kind of defeats the purpose as it is clearly less efficient.

I simply create a mapping for this type of enumerated attribute name in say excell, making it very as large as anticipated need.

When not enumerated I just copy the list from the feature information window and parse it using like notepad++.

You could replace the exposer with python attributecreation ass the data is enumerated.

I guess doing that would increase the dataset by 2.

This will go on till safe figure out how to blindly expose everything hidden.

If you are writing it out dynamically to excel, there's no need to expose the attributes in the workspace

@egomm

So I can dynamically write dynamically created attributes?

How would the schema be created?

Always keen to learn some new technique, I am.

pivot_example.fmwt An example @gio

@egomm

To get to the attributes I would have to read the file again and still not know what the attributes are, unless I sacrifice dynamism.

Exposure of enumerated attribute names makes it possible to access them in the workbench. (I simply create a long import list)

Not required for the solution of this topic I admit.

But for this topic probably adequate.

Thanks Takashi. This is yet another solution! I currently have an AttributeCreator built into a similar workflow rather than an AttributeRenamer.

I'm still probably going to take a crack at setting up egomm's workflow if I have time. I'll be pushing a number of disparate datasets with identical schema thru this workflow, and I never know how many Secondary_ID values there may be for a given ID... I suspect up to a few hundred for some of these datasets. I can set up the AttributeRenamer/Creator to accommodate this, but egomm's workflow is definitely more elegant.

Thanks all for your help. It's much appreciated.