Forum Discussion
Unpack .blob files in PowerBI
You should be able to apply Json.Document to the Contents column you are showing. Also, it looks like there are only a few types of files/data structures. You could disable load on the query you show, and then reference it in 2-3 other queries where you first filter using an identifying keyword before you apply the right Json.Document and navigation steps. You could then have these as separate tables in your model, or further transform them to match before you append them back together if appropriate.
Regards,
Pat
Hiya. Thanks for your help in solving this.
Apparently a large portion of the data rows are not in JSON-format, and it's independent of data-structure (I have found there are two basic ones). Weird since the attributes are similar for those it recognizes as JSON and those it does not. I guess it does not make sense to have half of the rows automatically divided in a separate table, if I still have to do it manually on the other half, and then stich it together. Any other suggestions?
(Another idea would be to ask the people providing the .blob data if they can nummerate the columns or change the structure to something more similar to regular .csv).
But I have made the Source-query, used Parse to Text (before I used Combine on the binaries), disabled load, referenced it from two separate queries based on the two different data structures and then manually used 6 - 10 x SplitColumns, then finally appended them together. The data is correctly presented in columns, but my concern is the speed. It's faster than before, where I used Combine rather than Parse to Text, but it still takes 5 hours to load 1.2GB data. Any suggestions on how to make it faster? Would it make it faster if I found a way to avoid all the SplitColumns and appending?
Hello Anonymous
as I already suggested you can try some "try otherwise" to try several data transformation in your TransformColumnns-function
BR
Jimmy
