Images in gt tables are not processed correctly by quarto into Typst #11829
Comments
|
I can repro this, thanks for the report. That's very strange. |
That's not surprising. GT emits HTML and Quarto needs to convert the table to Pandoc's AST format in order for it to have any chance to show up in Typst! |
|
GT (understandably) emits these images as data URIs; I wonder if this is fundamentally the issue. |
|
A markdown image encoded in a data URI does not cause Typst to crash: InputOutput
|
|
I did not attempt to test images inside gt when implementing Typst CSS / html table processing for Typst. It would be nice to have sparklines and such. Would think data URIs would work. |
|
They do work in a simplified setting outside of GT: That is a wholly-white PNG image, and this compiles to a table without a problem. |
|
Copy-pasting the image from the GT output into my simple table reintroduces the bug. So this is not GT: it's either Pandoc or Quarto. |
|
I can repro with your simplified example. On a hunch, I disabled juice, and both examples succeed. So it's either juice or my glue code around it. |
|
@gordonwoodhull It's a juice bug 😬 Instrumenting local function handle_raw_html_as_table(el)
local eltext
-- write el.text to disk
local f = io.open("/tmp/juice-input.html", "w")
f:write(el.text)
f:close()
if(_quarto.format.isTypstOutput()) then
eltext = juice(el.text)
local f = io.open("/tmp/juice-output.html", "w")
f:write(eltext)
f:close()
else
eltext = el.text
end
...Then we can already see the problem with the file sizes: |
|
It should be possible for us to protect data-URI-encoded images to work around |
|
Yeah, seeing the same thing. I think it could be a line length issue ( > 43300 characters) Agree, any kind of placeholder should work. Will try. |
|
Interestingly (?), calling |
|
Oh, that must be a bug they fixed in 11.0.0; we're running juice 10.0.0. EDIT: No, that's not it. Running juice directly from deno on those inputs doesn't trigger the bug (on version 10, or 11, either from |
|
Nod, I still see truncation if I change and version is 11 |
|
Okay I misread this at first: https://stackoverflow.com/questions/695151/data-protocol-url-size-limitations There isn't any fixed limit of 32K but it does seem that juice is deciding to truncate data URIs under 15K characters for its own reasons. |
|
Ok cool. Our workaround then is pretty clear - we intercept data URIs, replace them with UUIDs and then wrap it back. |
Bug description
I have found an issue while trying to use gt for tables with plots in Typst.
The plots that are created in the
text_transformfunction are copied into thefilename_files/mediabagfolder incorrectly. Some of all of them are broken and they appears incomplete. png and svg files seems truncated and jpg files have a broken encoding.I was able to reproduce the behaviour with the following gt functions:
ggplot_image,local_image,fmt_image.The issue may be related to Pandoc and the HTML processing that is performed on tables. However, when I tried using the
html-table-processing: noneoption the table simply did not appear.In the .typ output file it is clear that the issue is related to pointing the image to a broken copy of the picture, which happens even if a static image is provided.
Steps to reproduce
Expected behavior
Pictures that are copied into
filename_files/mediabagshould be identical to the originalActual behavior
Pictures that are copied into
filename_files/mediabagare broken, possibly truncated.Error message
Your environment
Quarto check output
Quarto 1.7.7 [✓] Checking environment information... Quarto cache location: /Users/filippofronza/Library/Caches/quarto [✓] Checking versions of quarto binary dependencies... Pandoc version 3.4.0: OK Dart Sass version 1.70.0: OK Deno version 1.46.3: OK Typst version 0.11.0: OK [✓] Checking versions of quarto dependencies......OK [✓] Checking Quarto installation......OK Version: 1.7.7 Path: /Applications/quarto/bin [✓] Checking tools....................OK TinyTeX: v2024.10 Chromium: (not installed) [✓] Checking LaTeX....................OK Using: TinyTex Path: /Users/filippofronza/Library/TinyTeX/bin/universal-darwin Version: 2024 [✓] Checking basic markdown render....OK [✓] Checking Python 3 installation....OK Version: 3.13.1 Path: /opt/homebrew/opt/[email protected]/bin/python3.13 Jupyter: 5.7.2 Kernels: python3 (|) Checking Jupyter engine render....Traceback (most recent call last): File "/Applications/quarto/share/jupyter/jupyter.py", line 21, in <module> from notebook import notebook_execute, RestartKernel File "/Applications/quarto/share/jupyter/notebook.py", line 15, in <module> from yaml import safe_load as parse_string ModuleNotFoundError: No module named 'yaml' [✓] Checking Jupyter engine render....OKNote: I am on the pre-release version right now because updated in the hope of resolving the issue. The issue was the same in the 1.6 version I was using before.
The text was updated successfully, but these errors were encountered: