Splitting data from excel cells

Hi @gisuser,

I would use the steps outlined by Takashi on the previous post you commented on here: https://knowledge.safe.com/questions/42016/hi-all-2.html

You'll then need to use an AttributeManager to clean up the attributes.

attributesplit.fmw

Hi @hollyatsafe

thank you, but the trick in this subject that we have in the second column data in a different language (Arabic),

then based on your test A it merged with "1", but it should be merged with " 3".

English characters "L_R", but Arabic is "R_L"

and that what Attributesplitter do split from L to R which make the join based on the index id difficult

Hi @gisuser,

Ah my mistake, ok well after the ListExploder for the Arabic characters we can add a Sorter to revert the order of _element_index using Numeric / Descending, then add a Counter to replace the _element_index with the correct count.

See the revised workspace to demonstrate this: attributesplitrevised.fmw

Thank u so much @hollyatsafe, but what if the case was more complicated like the below table, we have thousands of rows.

????????

UAE

?????, ???, ????

Lebanon, Egypt, Without

????????

UAE

????

Without

????????, ?????, ????

UAE, Lebanon, Without

????????, ??????, ?????

UAE, Jordan, Lebanon

????????

UAE

The attribute splitter for me splits right to left with arabic characters

So the following process should work

Use two attribute splitters, to split both columns, make sure to give the lists different names. Explode one of the lists, and then use the attribute element index to in an attribute creator to access the correct list element for the other column

Thank you @ebygomm, that was so workable and helpful, but what if the case was the Arabic letter for the English was written from lift to right like the below :

R1

A, B, C

?, ?, ?

R2

A, B, C

?, ?, ?

There is a custom transformer 'ListReverser' that you could use to reverse the list order, prior to using the attribute creator

it is ll be workable if we have just one row of the data but when you have multi rows it won't be workable

I'm not sure why you think the number of rows of data matter. Reversing the order of the list happens for each feature individually, exactly as required

If there is the following situation where Column1 should be left to right and Column 2 should be right to left

You can say that column2 needs to be reversed and get this output

Hi @gisuser,

Having reviewed the suggestion by ebygomm I think that is the simpler solution, however I have also updated my workspace if you'd prefer to go that route. Given a more complex dataset we cannot use the element_index to join as these values will be duplicated for each list. Instead we can create a new attribute to join/aggregate on that we can ensure will be unique for each feature.

attributesplitrevisedv2.fmw