I have a file that will not split when I import it into Scriver (Windows v3). It was originally a google doc that I downloaded as a .docx file, then opened in Libre Office and re-save as a .docx file. All the heading styles are there (I only used H1 and H2). Every time I import it telling scrivener to “Split using documents outline structure” it just imports the whole document as a single section. The ironic part is that the heading styles are preserved in the text!
I am very confused about why this is not working, when scrivener clearly “sees” the headings. Going in and splitting the section up manually is possible but I had been hoping to avoid the extra work.
I would try saving to Libre, but as an rtf file (Scrivener’s native language) check the styles panel in Libre to verify the appropriate headings are applied, and then import with splits. (Caveat - I have never used google docs, but I do know that Libre and RTF both play well with Scrivener.)
The OP’s post was from a while ago, so this is more for future reference.
Something to check in LibreOffice is whether there is a navigation outline to begin with. You will find this by opening the sidebar (Ctrl+F5 by default I think), and then clicking on the compass icon. Under the “Headings” section at the very top of this sidebar you should see an outline that corresponds with the heading structure. If you do not see that, then the stylesheet saved from Google Docs does not have an outline. Heading levels do not automatically imply an outline.
If a document is missing one, LO could be used to fix it, by modifying the levelled headings. For each heading level:
Right-click on the heading style in the sidebar and click “Modify”.
In the Outline & Lists tab, set the Outline Level to the level corresponding with the heading level.
After fixing all the heading levels you desire to have outlined, you should be able to switch back to the Navigator tab and see them listed in a nested hierarchy in the Headings area.
As far as results go, from a file edited in LibreOffice:
DOCX: best all around.
ODT: works fine so long as you elect to strip out headings during split (which is often desired anyway as Scrivener’s design prefers headings to be outline-driven rather than text editor driven; this is most likely a limitation in the third-party ODT converter we use, as I’m sure LO is saving to its own format correctly).
RTF: Levels 1 & 2 will import correctly if headings are stripped, but level 3 (and presumably greater, but that is all I tested) will come in as a bullet line. It is worth noting that LibreOffice’s RTF handling is sub-par in this area. If you save a properly outlined document to RTF and then reload it in LibreOffice, the heading structure will be lost. So we can’t know for sure if it is even saving it properly to begin with (though it seems to be doing so in part).
To clarify these are results from LO. Other word processors may have different peculiarities. As for why ODT and RTF would prefer stripping headings, it goes back to how LO uses a listing environment for headings. With DOCX this is not an issue, and you could have the headings imported as styled text. With ODT and RTF they will come in with bullets. Stripping the heading lines avoids that problem (at least up to level 2 for RTF).