Static web site from MS Word/LO Writer files?

[Shohreh]

Hello,

Searching this section with "word web site " returned no hit(s).

For a non-techie friend who just retired from university as an economics professor and would still like to publish academic articles, I need to find a simple way to create a web site.

A simple web server (Linux + Nginx) running on a thin client is all it's needed in terms of hardware. As for software, static web pages are plenty good, no need to bother with dynamic solutions like Wordpress et al.

As for publishing articles, he's used to writing them in Word (including graphs, that are impossible/too hard to draw in HTML), and them export them to PDF.

I'd be surprised if there was no solution that…

1. Turns Word files into HTML and PDF
2. Creates a new article as HTML with the PDF as attachment for those needing that instead (printing, visual problems)
3. Updates the site's homepage accordingly

It looks like Pandoc can turn Word files into HTML and PDF. Is there a solution that could then (re)generate the web site every time an article is added/edited?

Thank you.

[iandol]

Quarto has a nice setup for publishing web pages hosted on several existing platforms, and can generate static HTML too. It uses pandoc markdown as its engine, so you could add in a script to auto-convert from docx to md files into a git repo and then automate quarto for publishing...

https://quarto.org
https://quarto.org/docs/websites/
https://quarto.org/docs/gallery/#websites

[Shohreh]

Thanks for the tip. Is the docx to md, and then md to HTML conversion reliable?

[iandol]

This totally depends on what Word document features you are using. Honestly, using Word as the source is not ideal, using a markdown frontend would be much better; many markdown writing apps, including free ones offer a much cleaner writing experience than Word, and they will facilitate the "cleanest" maintenance of source to final target. The best way to see how it works is to make a minimal document that contains the features she/he will use, and try a manual conversion to markdown, see what the markdown looks like to give you an idea what makes it across from Word.

There are some conversions things like custom styles https://pandoc.org/MANUAL.html#input that you need to manually turn on when converting from docx, and specific things like numbered figure captions, possibly native citations etc. will need looking at. Some people suggest sticking to strict markdown when converting from DOCX: https://gist.github.com/plembo/409a8d7b1bae66622dbcd26337bbb185 -- but try both "full" and "strict" to see what works for you.

---

On the website end, Quarto is helpful, you can organise the documents and links how you please; and while noting there are other tools that orchestrate pandoc more directly, like pandocomatic, I think Quarto is beter overall.

[Shohreh]

Thanks. I know Word isn't necessarily a very good tool to source articles, but like quite a lot of people out there, he's used to it to write academic articles. I'm surprised no tool seems to exist for such a common need (Word → HTML → static web site, either at home or hosted.)

There's no way he'll move to md. I'll see if LibreOffice is a better solution (on the Mac) as a source for eg. pandoc.

[iandol]

I suspect Word will be more flexible as it supports reading custom styles which ODT input does not. I took this small doc in both ODT and DOCX format -- it contains character style (emphasis style in the first occurence of "Text"), direct formatting (bold), a figure with caption, headings and a maths equation:

test.docx
test.odt

And got these outputs:

odt -- pandoc -s -t markdown -f odt test.odt

Test Title

A Test

# Abstract

Some *Text* here. Text with **direct formatting**.

![This is a
caption.](Pictures/1000000000000B89000003CA2C349683.png){alt="Figure 1: This is a caption."
width="17.59cm" height="5.777cm"}

More text.

$$x = {y + 4}$$

# Discussion

## Sun

Some text.

## Moon

Some text.

docx -- pandoc -s -t markdown -f docx test.docx

---
subtitle: A Test
title: Test Title
---

# Abstract

Some *Text* here. Text with **direct formatting**.

![This is a caption](media/image1.png){width="6.925in"
height="2.2694444444444444in"}

More text.

$$x = y + 4$$

# Discussion

## Sun

Some text.

## Moon

Some text.

docx+styles pandoc -s -t markdown -f docx+styles test.docx

---
subtitle: A Test
title: Test Title
---

# Abstract

::: {custom-style="Body Text"}
Some [Text]{custom-style="Emphasis"} here. Text with **direct
formatting**.
:::

<figure data-custom-style="Body Text">
<img src="media/image1.png" style="width:6.925in;height:2.26944in" />
<figcaption><div data-custom-style="caption">
<p>This is a caption</p>
</div></figcaption>
</figure>

::: {custom-style="Body Text"}
More text.
:::

::: {custom-style="Body Text"}
$$x = y + 4$$
:::

# Discussion

## Sun

::: {custom-style="Body Text"}
Some text.
:::

## Moon

::: {custom-style="Body Text"}
Some text.
:::

docx strict markdown -- pandoc -s -t markdown_strict -f docx test.docx

# Abstract

Some *Text* here. Text with **direct formatting**.

<figure>
<img src="media/image1.png" style="width:6.925in;height:2.26944in" />
<figcaption><p>This is a caption</p></figcaption>
</figure>

More text.

*x* = *y* + 4

# Discussion

## Sun

Some text.

## Moon

Some text.

At least figures, captions, equations, headings and inline emphasis or direct bold formatting works across odt and docx -- docx additionally seems to convert title and subtitle styles to metadata, and you can use custom styles which you can then hook into custom CSS transformations for each block.

[Shohreh]

Thanks very much.

Alternatively, maybe the tool could just read the metadata from Word to fill the title and a basic body in an HTML article, and attach the PDF with the full article before regenerating the home page.

[iandol]

PDFs have low accessibility. I think using Quarto to convert the markdown you generate from docx files to static HTML with navigation among articles is a much better solution. Quarto can even provide a PDF/ODT version of a static HTML page for readers to download their own copy.

The DOCX>MD>HTML pipeline can be fully automated via a script, and if you are competent you can use continuous integration (CI) with github etc. so the web site auto-rebuilds based on a new files uploaded etc..

https://quarto.org/docs/publishing/quarto-pub.html#github-action

If you don't know how to handle CI, then a local script that triggers on a new docx in a folder, which converts docx>markdown, then calls quarto to generate the site then upload it somewhere eould make this low maintenance