fix: preserve 'rmd' code chunks in Rd

NEWS.md

@@ -1,5 +1,6 @@
 # pkgdown (development version)
 
+* Preserve Markdown code blocks with class rmd from roxygen2 docs (@salim-b, #2298).
 * Avoid unwanted linebreaks from parsing `DESCRIPTION` (@salim-b, #2247).
 * Remove redundant entries in the documentation index when multiple explicit `@usage` tags are provided (@klmr, #2302)
 * The article index now sorts vignettes and non-vignette articles alphabetically by their filename (literally, their `basename()`), by default (@jennybc, #2253).

R/tweak-reference.R

@@ -61,6 +61,8 @@ tweak_highlight_other <- function(div) {
   lang <- sub("sourceCode ", "", xml2::xml_attr(div, "class"))
   # since roxygen 7.2.0 generic code blocks have sourceCode with no lang
   if (!is.na(lang) && lang == "sourceCode") lang <- "r"
+  # Pandoc does not recognize rmd as a language :-)
+  if (tolower(lang) %in% c("rmd", "qmd")) lang <- "markdown"
   # many backticks to account for possible nested code blocks
   # like a Markdown code block with code chunks inside
   md <- paste0("``````", lang, "\n", xml2::xml_text(code), "\n``````")

tests/testthat/test-tweak-reference.R

@@ -112,6 +112,14 @@ test_that("can highlight other languages", {
   expect_equal(xpath_text(html, "//code//span[not(span)]")[[1]], "field")
 })
 
+test_that("can highlight 'rmd'", {
+  skip_if_no_pandoc("2.16")
+  html <- xml2::read_xml('<div class="rmd"><pre><code>field: value</code></pre></div>')
+  tweak_highlight_other(html)
+
+  expect_equal(xpath_attr(html, "//code//span[not(span)]", "class")[[1]], "an")
+})
+
 test_that("fails cleanly", {
   html <- xml2::read_xml('<div><pre><code></code></pre></div>')
   tweak_highlight_other(html)