In my free time I’ve been learning about the XML format of MS Word documents. And if you’re curious, here’s my advice for how to learn about Word’s XML format: don’t Thank you for coming to my Ted talk. - ThreadSky

jdlong.cerebralmastication.com • 85 days ago

In my free time I’ve been learning about the XML format of MS Word documents. And if you’re curious, here’s my advice for how to learn about Word’s XML format: don’t

Thank you for coming to my Ted talk.

Comments

jeremy-data.bsky.social•85 days ago

Thanks for just stating your conclusion and not making it a 200 page book

jdlong.cerebralmastication.com•85 days ago

Not all heroes wear capes.

christophscheuch.bsky.social•85 days ago

On the other hand, if the book were written in XML, it‘d make a strong argument

id10terror.bsky.social•85 days ago

Once used Excel + Azure Functions to automate cloud based processing. Used the Closed XML library to pull in Excel and parse the instructions for predefined processes. The idea was a to use a tool familiar to business users for automation.

https://github.com/ClosedXML/ClosedXML

mrtimnewman.bsky.social•85 days ago

Key design principle: sadism

christiannolan.bsky.social•85 days ago

So annoying when something uses standards but .. doesn’t let you work with them very well!

jamesgoldie.dev•85 days ago

Learning the Office implementation of OpenXML made me absolutely miserable 😮‍💨

milesmcbain.bsky.social•85 days ago

I once went on the same journey with Excel. It’s just xml how hard can it oh no

jdlong.cerebralmastication.com•85 days ago

Sorry for your loss. 💔

jdlong.cerebralmastication.com•85 days ago

The mistake was mine, tbh. I mistakenly thought that I could parse the plain text and make alterations the put it all back together without MS Word caring. It turns out one really has to parse the whole thing. Open source libs can’t access all elements. Altering the XML directly is very error prone

topepo.bsky.social•85 days ago

I read that damn document when I was making odfWeave. APIs with names like ‘FornatLikeWord95’. Then draw the 🦉

The OSF docs were much better.

Thankfully knitr came along and I could orphan odfWeave.

bk24five.bsky.social•85 days ago

At the risk of sounding stupid maybe try using an html parsing library due it being simular syntax and html library's being more robust. Just a thought.

jdlong.cerebralmastication.com•85 days ago

It’s much worse than html 😂

bk24five.bsky.social•85 days ago

Never tried it myself but because html and xml are both markup languages my assumption would be any html parsing library that can parse an html document should also be able to parse an xml document. Just a though I may be wrong!!

jdlong.cerebralmastication.com•85 days ago

Parsing the markup is trivial. Doing anything useful with the results is not.

geoffjentry.bsky.social•85 days ago

My recollection was the while thing was a 🖕to a growing group of would be regulators demanding they have an “open” format. And it was, but it was impossible to do anything with it

jdlong.cerebralmastication.com•85 days ago

Yeah I think that’s right.

jdlong.cerebralmastication.com•85 days ago

You can buy a library that can parse everything in a word doc including text boxes. At $2k these are a steal. Save yourself the pain

doriantaylor.com•85 days ago

also unoconv (which has bindings) will turn docx (dogshit) into odt (much cleaner)

jdlong.cerebralmastication.com•85 days ago

Most of these tools end up breaking some formatting. I’ll try uniconv though.

doriantaylor.com•85 days ago

unoconv is just headless star/open/libreoffice

doriantaylor.com•85 days ago

oh huh they made it into a server https://github.com/unoconv/unoserver/

mark-nenadov.bsky.social•85 days ago

Appreciate the PSA (Pain-saving Service Announcement?). I have no experience with this particular thing but even if you were the least trustworthy person, what you say rings with plausibility 😅

shinyblackshoe.bsky.social•85 days ago

R has some libraries that manage this reasonably.

But it likely does depend exactly what you are doing.

Replacing colour with color - probably fine.

Replacing paragraph 1 with 18 paragraphs and 6 figures...

...it's gonna break

jdlong.cerebralmastication.com•85 days ago

“Replace text in a text box” - all open source libraries fail 😞

shinyblackshoe.bsky.social•85 days ago

Box size changes at a guess?

jdlong.cerebralmastication.com•85 days ago

Boxes are hidden in weird content that’s not “paragraph content” and it’s apparently hard to work with. Lots of discussion in GitHub repos that basically end with “we’re just not going to mess with that”

shinyblackshoe.bsky.social•85 days ago

https://stackoverflow.com/questions/64211609/adding-image-to-text-box-using-officer

This claims to be swapping text and inserting images into text boxes.

But as I say - if the box size changes... It will end in tears. It often. Does in word!

gdeejay.bsky.social•85 days ago

Solid advice.

noamross.net•85 days ago

I have been and am once again deep into this. It's an exercise in sin-eating.

jdlong.cerebralmastication.com•85 days ago

Tots and pears.

marcoshuerta.com•85 days ago

If recall correctly, when it first came out, the MacOS version of Word couldn’t open .docx; then a second converter app was needed for months (if not longer.)

jdlong.cerebralmastication.com•85 days ago

That’s very believable

bjbloom.bsky.social•85 days ago

“Nobody likes PDFs as output of R Markdown. Maybe I can see if I can get it to work in Word.” Thoughts of many R programmers who immediately regret the hours spent trying to figure this out. Cousin to the thought “let’s just put this excel file into Gitlab.” The MS wormhole of hell.

bjbloom.bsky.social•85 days ago

Comments

Posting Rules

Reply