quite a few "... and claude" commits lately at least.
AussieWog93 48 seconds ago [-]
There's a genuine difference between pure vibe coding and just using Claude to polish things a bit.
dmd 2 hours ago [-]
So I just tried this with a bunch of medium-complex documents and it's wildly wrong. I suspect the authors have never seen an actually complicated Word document?
bulder 4 hours ago [-]
The post title about it being "pixel-faithful" is a bit strange. I don't see that claim in the repo, and they don't seem to even claim full feature support at the moment. And for the features marked as supported in .pptx's, it does seem that at least slide image backgrounds and bullet point images aren't actually working, and some text objects have inverted text colors. Seems quite far away from being pixel-faithful in fact.
Both the GitHub repository and the website explicitly claim the entire project is entirely generated by Claude.
victorkulla 9 minutes ago [-]
Apparently "Hacker News" won't allow me to delete my account. Well let's give this a go:
HACKER NEWS is ran by a bunch of idiots that think karma is an actual thing, If this is so - then explain this?
jsmilker 1 hours ago [-]
I ran it against some of our internal test files, and it failed on all of them how was this project even tested for proper compliance
userbinator 1 hours ago [-]
No human-written application code exists in this repository.
It's 100% hallucinated.
jbgt 4 hours ago [-]
Interesting because I'm building ooxml-cli right now, for editing pptx, docx, xlsx.
At work I had to adapt a pptx to a corporate template and tried via agent. It kept failing so I started building and then it was able to relatively quick and accurate do what I needed.
Then I needed it to make tables, add pictures. Recently wanted to get data from an xslx and replace text in a presentation etc.
So the tool is growing and maybe this would be interesting to have as the non LibreOffice dependent viewer...
Misread that as open office xml not office open xml. I wish the standards were named more differently. They are too easy to confuse
stuaxo 4 hours ago [-]
Microsoft did that deliberately.
lovasoa 3 hours ago [-]
I tried a few PPTX files from consulting firms that are available online. The rendering does not seem to be truly pixel-perfect, but all of them were quite readable and had a good layout, which is already an impressive feat.
wis 4 hours ago [-]
Very nice, the rendered demo for all the file types appear to render flawlessly and load instantly on page load, and looking in the DevTools the parsers are split into different Wasm bundles for each file type xslx, docx and pptx:
docx 458kb raw 217kb gzipped
pptx 574kb raw 253kb gzipped
xslx 601kb raw 269kb gzipped
I expected the Wasm bundles to be large and a lot more bigger than that for some reason.
ChatGPT.com can benefit from using this library (or such a library) for rendering a preview of the file in a side panel on the right, instead of just giving me a download link to the outputted/transformed docx/pptx/xslx file.
Jaxkr 2 hours ago [-]
I literally had to solve the "preview Office files in the browser" problem last week. I couldn't find a decent solution, so I ended up making a endpoint that ran the files through headless libreoffice on the server to convert them to PDF.
For PPTX and DOCX, this solution is slightly worse than libreoffice conversion (this does not appear to output highlightable text, while PDF conversion does).
However, the XLSX preview BLEW my mind considering this was AI coded. Really good, even interactive!
maxloh 2 hours ago [-]
> ...this does not appear to output highlightable text
I can't highlight text in Safari or Firefox on my iPhone (iOS 26.5), at least on that first page.
I'm not familiar with this application, so perhaps I'm missing a step, and editing mode.
vlmutolo 4 hours ago [-]
Pretty cool, rendering PowerPoint files to an image is probably the only way for LLMs to make sense of them.
Does this work in Cloudflare’s workerd environment? Would be nice to have a cheap serverless render -> LLM (GLM-OCR / PaddleOCR) -> Markdown pipeline for the various MS Office formats.
wmf 3 hours ago [-]
This code creates a JSON intermediate representation that LLMs could probably consume. You might want to simplify it to focus on content and reduce token usage.
gigel82 4 hours ago [-]
If someone actually got "pixel-faithful" Office documents rendering correctly, MS would be screwed. That's actually really important for a lot of companies that carry around decades-old templates that never look exactly right in LibreOffice or any other software that attempted to replicate it.
The slightest misalignment of a paragraph means a line on page 27 of 120 now moved down by 2 pixels, screwing everything else out of alignment. Yes, plenty of companies pay Microsoft 365 subscriptions because of exactly this reason; it sounds ludicrous when you think they could just pay someone to replicate the formatting in a different suite a lot less than the subscription costs, but that's not how it works...
quag 3 hours ago [-]
Sadly, Microsoft 365 is not “pixel perfect” compared to word. I often run into headaches where line numbers are different between the two and content ends up on different pages.
If Microsoft can’t get consistent rendering of word docs between Word for Windows, Word for macOS and Office 365, I don’t like anyone else’s chances.
ale42 2 hours ago [-]
Can the same version of Word now produce the same rendering on two PCs? In the past (I didn't really check recently, I'm thinking more about 10 years ago) the same file might have had page breaks in different positions and things like that. I never understood if it came from slightly different versions of the fonts, from some info derived from the default printing or anything else...
maxloh 3 hours ago [-]
I've heard that the same file can render differently in MS Word across different machines and OSs. So, that won’t help either.
phplovesong 4 hours ago [-]
Its kind of sad that the first thing in the repo is a mention that no human was involved in the programming.
userbinator 1 hours ago [-]
As others here have already mentioned, it doesn't work all that well either, proving that AI can't replace humans completely.
dizhn 3 hours ago [-]
Total "Damn if I do, damn if I don't" situation . I put a similar disclaimer on my AI stuff too. It would be much easier if they didn't mention it. But if they're like me, they want to give people the means for an informed decision. We can respect that and move on. Do we even know if actual big software companies aren't doing it?
llukas 4 hours ago [-]
Would author be able to do it otherwise? Is particular tool choice making result worse?
Bit identical/pixel-faithful reproductions are easy to verify…
applfanboysbgon 30 minutes ago [-]
> Is particular tool choice making result worse?
Well, yes, because it doesn't work.
> Bit identical/pixel-faithful reproductions are easy to verify…
And yet the prompter put so little effort in they couldn't even verify the software they prompted for does what it's supposed to.
NewsaHackO 16 minutes ago [-]
Why do you say it doesn't work? It seems to work fine for the feature parity they claim.
jstanley 4 hours ago [-]
"LLMs are amazing, I'm so much more productive now"
"oh yeah? Show me what you made, you can't, nobody can, it's all just AI psychosis"
"I made a pixel perfect Office document viewer"
"well... I wish you hadn't"
NetOpWibby 4 hours ago [-]
“If you use LLMs, you’re not a real developer, you’re lazy.”
which means it probably gets all the halucinated assets correctly and any real world documents wrong.
Still, looks pretty; if it actually has proper testing, could close the gap. Code not being the hard part is a major impediment to good software coming out of these things.
sublinear 5 hours ago [-]
> office-open-xml-viewer
alex_suzuki 4 hours ago [-]
[dead]
freakynit 4 hours ago [-]
"Built entirely by Claude through iterative prompting."
Holy cow!!
Rendered at 23:05:53 GMT+0000 (Coordinated Universal Time) with Vercel.
* https://github.com/eigenpal/docx-editor
** https://news.ycombinator.com/item?id=48228411
HACKER NEWS is ran by a bunch of idiots that think karma is an actual thing, If this is so - then explain this?
It's 100% hallucinated.
So the tool is growing and maybe this would be interesting to have as the non LibreOffice dependent viewer...
ChatGPT.com can benefit from using this library (or such a library) for rendering a preview of the file in a side panel on the right, instead of just giving me a download link to the outputted/transformed docx/pptx/xslx file.
For PPTX and DOCX, this solution is slightly worse than libreoffice conversion (this does not appear to output highlightable text, while PDF conversion does).
However, the XLSX preview BLEW my mind considering this was AI coded. Really good, even interactive!
Yeah, it does.
https://ooxml.silurus.dev/storybook/?path=/story/docxviewer-...
I'm not familiar with this application, so perhaps I'm missing a step, and editing mode.
Does this work in Cloudflare’s workerd environment? Would be nice to have a cheap serverless render -> LLM (GLM-OCR / PaddleOCR) -> Markdown pipeline for the various MS Office formats.
The slightest misalignment of a paragraph means a line on page 27 of 120 now moved down by 2 pixels, screwing everything else out of alignment. Yes, plenty of companies pay Microsoft 365 subscriptions because of exactly this reason; it sounds ludicrous when you think they could just pay someone to replicate the formatting in a different suite a lot less than the subscription costs, but that's not how it works...
If Microsoft can’t get consistent rendering of word docs between Word for Windows, Word for macOS and Office 365, I don’t like anyone else’s chances.
Bit identical/pixel-faithful reproductions are easy to verify…
Well, yes, because it doesn't work.
> Bit identical/pixel-faithful reproductions are easy to verify…
And yet the prompter put so little effort in they couldn't even verify the software they prompted for does what it's supposed to.
"oh yeah? Show me what you made, you can't, nobody can, it's all just AI psychosis"
"I made a pixel perfect Office document viewer"
"well... I wish you hadn't"
The best developers are lazy.
Still, looks pretty; if it actually has proper testing, could close the gap. Code not being the hard part is a major impediment to good software coming out of these things.
Holy cow!!