Using `gt::md()` renders different markdown syntax differently, depending upon where it is called.

[rich-iannone]

Using gt::md() renders different markdown syntax differently, depending upon where it is called.

Reprex:

---
format:
   docx: default
---


```{r}
gt::gt(gt::sp500[1:10, ]) |>
  gt::tab_header(gt::md("**Test header** <br>*__Test header two__*")) |>
  gt::tab_footnote(gt::md("_Test footnote_ <br>*__Test footnote two__*"))

Note:

• The header respects the bold and italics, but not the line break
• The footer does not respect any of the markdown syntax

Versions:

• gt: 0.11.0.9000
• quarto: 1.5.57

Originally posted by @MikeJohnPage in #1067 (comment)

[rjazwiec]

Will add here because my issue is very similar.

code:

 tab_spanner(., label = md("t~max~ [h]"), columns = matches("tmax")) %>% 
  tab_spanner(., label = md("C~max~ [ng/mL]"), columns = matches("cmax")) %>%
  tab_spanner(., label = md("t~1/2~ [h]"), columns = matches("t1_2")) %>%
  tab_spanner(., label = md("AUC~0-24h~ [h×ng/mL]"), columns = matches("auc_0_24h"))

generates output like this:

when there are only numbers and letters in the subscript part everything renders properly:

Any other sign between ~ makes it render strikethrough instead of subscript (vulgar ½ doesn't work either)

It works like this in pdf, png, html. In docx output it doesn't render subscript in tab_spanner at all just puts Cmax as it is with tildas visible

Best regards
Radek

ps. I just saw that GH has done the same with Cmax above. max is typed between ~ (one per side not two) and is rendered as striketrough.

[rjazwiec]

Also md("AUC$_{0-24h}$ [h×ng/mL]") renders properly.

[olivroy]

The problem really seems to be simply limited to footnotes if I am not mistaken.

I finally figured it out!

It seems that markdown is processed at build stage for source notes.

dt_source_notes_build() calls process_text() which resolves the markdown.

Therefore, when we reach the create_source_notes_component_xml() step in the render process, we don't need to process Markdown.

So, in other formats, the grid for example, we need to call process_text() in create_footnotes_component_g() to make sure that markdown is correctly processed. idem in create_footnotes_component_h().

Edit: line break now works in the dev version, thanks to #1983