I have a thought that it might have helped resolve a missing delimiter in the file. What happens if you add the column, save the file, then remove the column and reconnect with Tableau Desktop?
Thanks for offering your help.
I used excel to remove the "padding" column and re-saved the CSV.
There are two (still wrong) behaviors of tableau reading this new CSV:
- If refreshing the data from an existing tableau which originally saw the CSV with the "padding" column tableau holds the data types but some wrong data moves across dimensions
- If creating a new data source from the save CSV I see the original problems but with different wrong data moving across dimensions (compared to the situation above)
Puzzled by this and expect it's my ignorance on what tableau does behind the scenes.
PS all CSV files are saved with a "UTF-8" text format
When opening the CSV file in Excel there are no malformed rows appearing? Only within Tableau Desktop?
That is a very interesting behavior that I might suggest opening a case with the Tableau Support team to have a deeper look at.
Hope this helps!
That's right, opening & viewing via excel does not yield the strange behavior. I'll open one up then.
Many thanks for your help Patrick.
This doesn't directly answer your question but I wanted to throw out that that I've been developing a package that allows you to drop a pandas DataFrame directly to a Hyper extract. If you are using Tableau 10.5+ it may be of value to you, so feel free to give it a look and let me know of any questions!
Tim, did you ever get any feedback/resolve for this issue?
Spotted this a few weeks ago myself, general strange behavior with csv files such as:
fields belonging to one field appearing in the column for another field (not all the data for that field, most was correct, just a handful of rows)
some fields ended up with a ' ' in front. E.g. [Country] could have a [UK] string and a [ UK] string
really no idea what the cause of this is. Seems to be reading the data wrong when looking at it here:
as well as in the worksheet itself.
In the csv file however, no such problems. All data there is as expected.