Specifying tab as separator in "CSV" resource file for Cloud

I have a TSV (Tab Separated File) I'm trying to process using Designer with Stardog Cloud.
I am able to browse to and open the file once I renamed the extension to .csv (I could select the file when named .tsv but it was then ignored).
The next screen allows me to specify the separator character, however I can find no way to specify it as tab - I have tried \t, TAB, 0x09, 9 and all variants.
Other Stardog docs say to use \t but I'm guessing the Designer UI is not recognizing it and passing it through. Is there a secret to this or any workround?

Hi Pete,

In the Designer UI, only pasting a tab into the separate character box is supported. To copy a tab character, I usually open a text file, enter a tab, copy, and then I can paste the tab character into Designer.

I have shared the feature request to be able to define tabs more easily as separators.


Great, that worked, thanks for quick response, Laura.


I spoke too soon. I was able to do a simple Mapping for my TSV file, but when I then try to "Publish" the mapping and data it wants me to re-upload the (large) file. However when I select the file I get the error "The headers in ... do not match those of the original file". I guess it needs to be told again to use tab as the separator but I cannot find a way to specify that.

A further issue - the Publish seems to insist on a named graph as the target. How do I tell it to use the default graph?


Hi Pete -- sorry about that. You are correct, while pasting a tab character works for initially parsing a file, it is not supported when publishing. I would recommend using a comma or another delimiter that is a typed character, such as a pipe (|).

We don't support publishing to any special named graph, including the default graph from Designer. Can you expand on why you want to publish your data to the default (no context) graph?

Even if I used pipe, how would I tell the publishing step to use that instead of comma? Wouldn't I get the same problem I'm having now with tab?
So is the ability for Designer to specify separators completely pointless as part of the workflow when using Stardog Cloud?

The problem with comma is that certain columns in my data files have multiple values within the column separated by comma.
Looks like I need a whole data prep step to replace those commas with something else and then tab by comma. Or put quotes around the comma-separated values. Not hugely hard but not so trivial since these are large files I'm dealing with and updated frequently. I thought this data prep was what Designer was supposed to be for!

Regarding the default graph, I would normally use named graphs, but I'd had some hassle a while back when preloading and querying some data via Studio - though it seemed I had loaded data in to a named graph via Load Data, there was no way to query it: as I recall, even general queries such as SELECT DISTINCT ?g WHERE {GRAPH ?g {?s ?p ?o}} returned nothing, as well as using the specific graph URI in FROM or FROM NAMED. Instead of fighting that I reverted to using the default. I guess I need to go back and fight that battle.

Thanks for help so far,

Using a pipe as the delimited works without re-specifying the delimiter when uploading the large file during publishing. A comma, pipe, or another character typed into the delimiter field can be used. I'm sorry for any confusion caused by suggesting a workaround with the pasted tab.

Thank you for expanding on the desire to use the default graph. If you continue to have issues querying for data in specific graphs, it would be great to better understand better the steps to reproduce.