Test Scrivener's New .docx Converters

KB · July 1, 2019, 1:19pm

jpkell:

Hi Keith,

I am a longtime Scrivener user, who is now writing a scholarly book with it. The book has many equations and symbols, for which I used MathType integrated with Scrivener. But I have encountered a huge problem that I think your new converters could help with (and from reading the forums, I’m not the only one who could use this).

When one’s manuscript contains equations and symbols, it turns out that publishers want either a LaTex document or a Word document with equations and symbols editable with Word’s Equation Editor. But when one compiles a Scrivener document that includes MathType equations and symbols to docx, the equations and symbols become images that are not editable with Equation Editor.

After spending a day yesterday trying to figure out a workaround, the best I can think of is to (re)write the book in Markdown with Scrivener, inserting where necessary the LaTex equation/symbol code that I can copy from MathType. I confirmed that I can use pandoc to convert a markdown file containing LaTex equation code to a docx file containing an equation that is editable in Word’s Equation Editor. So I know it’s possible to convert the LaTex equation syntax to Word’s Equation Editor. What would be AMAZING-AND-I-WOULD-BE-FOREVER-GRATEFUL is if I could write in Scrivener as usual, with all of Scrivener’s usual rich text, but instead of inserting MathType equations as images, I would insert them as LaTex code, and then your tools would use Pandoc (or equivalent) to render the equations as editable with Equation Editor when compiling to docx from within Scrivener.

Is there any chance at all this could be included in your new docx converter tools?? Thanks!

I don’t really see how this would be very feasible, I’m afraid (it requires a Pandoc install for a start, and involves mixing up different types of export). MathType equations contain MathML, but there’s no easy way of converting MathML to the OMML that Docx files use.

All the best,
Keith

nontroppo · July 2, 2019, 1:22am

Hi KB, my screenshot in this thread shows the bug, the screenshot in the other thread shows how it should compile : what you should see in Word is THREE Level 1 headings « Part One » « Part Two » « Part Three » with chapters as level two, but your Word doc shows only the first « Part One » Level 1 heading present (well it does in my Word V16.26), if you go to « Part Two » text in Word you’ll see it is a normal style, explaining why it doesn’t make it to the outline.

Here is the Outline in Scrivener:

Here is a screenshot of what it is supposed to look like in Word:

gr · July 5, 2019, 12:44pm

The following tech support thread is about a problem someone had with getting footers to work and which was resolved by turning ON enhanced converters. It might point to a kink in the in-house converters, so just dropping a note here in case.

[url=https://forum.literatureandlatte.com/t/header-footer-not-exporting-to-word-on-compile/46106/1]

gr

rayl · July 15, 2019, 9:30pm

Footnotes and live links to email addresses seem to be lost if a .docx file is imported using the new converters, but they are retained when using the Aspose converters.

However, footnotes written in Scrivener and exported using the new converters seem to be fine.

(Using Scrivener 3.1.3 on macOS 10.14.5)

Ray

rayl · July 16, 2019, 3:18pm

Further to this, after experimenting with two more .docx files.

Embedded graphics objects - mostly tables and graphs pasted from Excel into Word- are not imported using the new converter, but are imported using the Aspose converter. This applies to all three files.

Footnotes are imported using the new converter for just one of the three .docx files.

Each file is the result of a collaboration, so some parts were originally drafted in Scrivener, but subsequently edited, including inserting the graphics objects, in Word for Mac and Word for Windows.

All best

Ray

KB · July 20, 2019, 5:36pm

Please post sample .docx files so that I can test these issues.
Thanks,
Keith

rayl · July 20, 2019, 9:25pm

Hi Keith

Herewith a zip file with 3 documents:

“Recent Work” is a short document created using Word for Mac 2011, with a few footnotes and embedded (by copy/paste) spreadsheet and chart. This imports perfectly with the new converters.

“RP paper extract” is a few pages from a collaboratively authored academic paper, with footnotes and embedded graphics, substantially drafted by my colleague using Word for Windows. The footnotes and graphics do not import with the new converter but are OK using the Aspose converter.

All best

Ray

Converteg.zip (272 KB)

jpkell · July 23, 2019, 7:05pm

Thanks very much for the reply. I have now decided to write the book in Scrivener with AmberV’s LaTeX template and compile to LaTeX, bypassing Word entirely. That seems like the best way forward.

JohnnyZMusic · August 2, 2019, 6:50pm

Will this convert and import multiple Pages documents?

kewms · August 2, 2019, 9:18pm

No. Apple has not published the Pages format, making it difficult to write converters for it.

We recommend using Word as the exchange format between Pages and Scrivener.

Katherine

mcl · August 22, 2019, 5:25pm

I’m posting this file here as a result of the tests on importing tables from Word conducted on this thread: https://forum.literatureandlatte.com/t/importing-tables-from-word/46719/1

Hope this is the right place for it.
table import test.scriv.zip (167 KB)

nontroppo · October 23, 2019, 10:45am

Just to mention, the Heading 1 outlining bug is still present in V3.1.4, the output shoud have three “Level 1” sections each [Part], with three “Level 2” sections [Chapter], each with two “Level 3” sections [Scene]. In the Scrivener converter output, “Level 1” Part Two and Part Three are missing:

(post with test project here).

nontroppo · November 18, 2020, 3:52am

This issue/bug is still present in V3.2, and though not relevant to Scrivener’s converter per se the Aspose output is now also missing level 1 and level 2 headings. That means at present the method of assigning Heading 1-6 styles to section titles to allow proper outlining in DOCX is currently broken in V3.2; unless there is a new way to achieve this.

KB · November 19, 2020, 8:18pm

Sorry, is this a bug in the Aspose conversion only, or in Scrivener’s own .docx converter too? I’m confused as this is the Scriv converter thread but you seem to be saying it’s the Aspose ones. If it’s Aspose, how does RTF do? (Aspose converts from RTF.)

nontroppo · November 19, 2020, 11:07pm

This is an issue in Scrivener’s converter, where only the first Level 1 heading gets assigned the “Heading 1” style (and thus Word cannot outline without a styled heading). This applies to V3.14, and remains an issue in V3.2. There is a regression in Scrivener V3.2 in that the RTF (and as a result the Aspose output) now does not get Level 1 or Level 2 headings, only Level 3 headings and you can see the result in the Word outliner in the screenshot here:

[attachment=0]Scrivener DOCX Bug.jpg[/attachment]

The Scrivener project is available here: https://forum.literatureandlatte.com/t/compiling-to-word-with-styles/39428/2

KB · November 20, 2020, 5:31pm

Thanks. I hope that I have now fixed both issues for 3.2.1.

nontroppo · November 21, 2020, 1:27am

I can confirm that both DOCX and RTF output now correctly exports all heading styles, thanks!

I am a bit confused by the new Preferences dialog:

[attachment=0]Screen Shot 2020-11-21 at 09.18.58_SML.png[/attachment]

As I understood it, there are 3 options: (1) macOS legacy converter. (2) Aspose converter, (3) Scrivener’s native converter. The menu gives two options, and I’m not sure how we can select between (2) and (3) using this dialog?

I forget now but I thought the older V3.1.x dialog also gave a bit more opinionated recommendation of which option is “better”? The new dialog doesn’t recommend one option over the other, though I think Scrivener’s converter is much better than the macOS one, while being faster than the Aspose one (though possibly less “complete”)… I almost wonder whether it is better to remove the macOS one (hiding it behind a defaults commandline setting or something)?

A small pedantic GUI text point, .doc is a different format than .docx; the dialog infers it only applies to the legacy .doc format, whereas it applies to both. I suspect most users don’t differentiate betweeen the legacy binary and “modern” XML formats (i.e. most users read .docx as .doc) and you are trying to simplify the description, but my geek pedanticism feels obliged to mention it

nontroppo · November 21, 2020, 2:03am

Ah, it seems I was using another JDK (openjdk that doesn’t symlink into /Library/Java/JavaVirtualMachines/) on my laptop, removing it and installing adoptopenjdk (which does symlink), and everything works:

brew uninstall openjdk
brew install adoptopenjdk

The standard installer is here: adoptopenjdk.net/?variant=openj … nt=hotspot

[attachment=0]Screen Shot 2020-11-21 at 10.01.35.png[/attachment]

This dialog only allows me to choose between (2) and (3) from the previous post, much simpler