feat: lint R Markdown out of the box

`.Rmd` is Markdown to the parser already: knitr chunk fences are fenced
code blocks, inline `r` expressions are code spans, and the YAML header
is frontmatter -- so a `[formats]` remap was all support ever took. The
extension is now mapped to Markdown itself, and a full R Markdown
document pins the behavior: chunks and inline code are never prose, and
everything else is.

Signed-off-by: Joseph Kato <joseph@jdkato.io>

Author: jdkato

internal/core/format.go

@@ -62,23 +62,23 @@ var FormatByExtension = map[string][]string{
 	`\.(?:adoc|asciidoc|asc)$`:                 {".adoc", "markup"},
 	`\.(?:clj|cljs|cljc|cljd)$`:                {".clj", "code"},
 	`\.(?:cpp|cc|c|cp|cxx|c\+\+|h|hpp|h\+\+)$`: {".cpp", "code"},
-	`\.(?:css)$`:                      {".css", "code"},
-	`\.(?:cs|csx)$`:                   {".c", "code"},
-	`\.(?:dita)$`:                     {".dita", "markup"},
-	`\.(?:go)$`:                       {".go", "code"},
-	`\.(?:hs)$`:                       {".hs", "code"},
-	`\.(?:html|htm|shtml|xhtml)$`:     {".html", "markup"},
-	`\.(?:java|bsh)$`:                 {".java", "code"},
-	`\.(?:jl)$`:                       {".jl", "code"},
-	`\.(?:js|jsx)$`:                   {".js", "code"},
-	`\.(?:lua)$`:                      {".lua", "code"},
-	`\.(?:md|mdown|markdown|markdn)$`: {".md", "markup"},
-	`\.(?:mdx)$`:                      {".mdx", "markup"},
-	`\.(?:org)$`:                      {".org", "markup"},
-	`\.(?:php)$`:                      {".php", "code"},
-	`\.(?:pl|pm|pod)$`:                {".r", "code"},
-	`\.(?:proto)$`:                    {".proto", "code"},
-	`\.(?:ps1|psm1|psd1)$`:            {".ps1", "code"},
+	`\.(?:css)$`:                             {".css", "code"},
+	`\.(?:cs|csx)$`:                          {".c", "code"},
+	`\.(?:dita)$`:                            {".dita", "markup"},
+	`\.(?:go)$`:                              {".go", "code"},
+	`\.(?:hs)$`:                              {".hs", "code"},
+	`\.(?:html|htm|shtml|xhtml)$`:            {".html", "markup"},
+	`\.(?:java|bsh)$`:                        {".java", "code"},
+	`\.(?:jl)$`:                              {".jl", "code"},
+	`\.(?:js|jsx)$`:                          {".js", "code"},
+	`\.(?:lua)$`:                             {".lua", "code"},
+	`\.(?:md|mdown|markdown|markdn|[Rr]md)$`: {".md", "markup"},
+	`\.(?:mdx)$`:                             {".mdx", "markup"},
+	`\.(?:org)$`:                             {".org", "markup"},
+	`\.(?:php)$`:                             {".php", "code"},
+	`\.(?:pl|pm|pod)$`:                       {".r", "code"},
+	`\.(?:proto)$`:                           {".proto", "code"},
+	`\.(?:ps1|psm1|psd1)$`:                   {".ps1", "code"},
 	`\.(?:rb|Gemfile|Rakefile|Brewfile|gemspec)$`: {".rb", "code"},
 	`\.(?:rs)$`:        {".rs", "code"},
 	`\.(?:rst|rest)$`:  {".rst", "markup"},

internal/core/util_test.go

@@ -13,6 +13,9 @@ func TestFormatFromExt(t *testing.T) {
 		".py":    {".py", "code"},
 		".cxx":   {".cpp", "code"},
 		".mdown": {".md", "markup"},
+		".Rmd":   {".md", "markup"},
+		".rmd":   {".md", "markup"},
+		".R":     {".r", "code"},
 	}
 	m := map[string]string{}
 	for ext, format := range extToFormat {

testdata/e2e/config-flags.yaml

@@ -87,7 +87,7 @@ cases:
       content/c.md:1:7:Vale.Spelling:Did you really mean 'ipsum'?
 
   - name: glob-negated-multi
-    args: "--glob=!**/*.{md,py,mdx} ."
+    args: "--glob=!**/*.{md,py,mdx,Rmd} ."
     exit: 0
     absent:
       - md

testdata/e2e/lint.yaml

@@ -191,6 +191,19 @@ cases:
       test.md:44:5:vale.Annotations:'TODO' left in text
       test.md:46:3:vale.Annotations:'TODO' left in text
 
+  - name: rmarkdown
+    about: .Rmd is Markdown out of the box -- knitr chunk fences and inline
+      `r` spans are code, and everything else lints as prose
+    args: test.Rmd
+    exit: 0
+    want: |
+      test.Rmd:19:47:vale.Annotations:'TODO' left in text
+      test.Rmd:35:25:vale.Annotations:'XXX' left in text
+      test.Rmd:52:23:vale.Annotations:'TODO' left in text
+      test.Rmd:58:32:vale.Annotations:'NOTE' left in text
+    absent:
+      - FIXME
+
   - name: assigned-ext
     about: --ext overrides the extension
     args: "--ext=.md test.xyz"

testdata/fixtures/formats/test.Rmd

@@ -0,0 +1,59 @@
+---
+title: "Quarterly Report"
+author: "Jane Doe"
+date: "2026-08-07"
+output:
+  html_document:
+    toc: true
+    fig_caption: true
+bibliography: refs.bib
+---
+
+```{r setup, include=FALSE}
+knitr::opts_chunk$set(echo = TRUE, warning = FALSE)
+library(ggplot2) # FIXME: only loaded for the appendix plots
+```
+
+# Overview
+
+This report summarizes the quarterly numbers, TODO confirm the cutoff date.
+
+We read the raw data and clean it:
+
+```{r load-data, message=FALSE}
+raw <- read.csv("data/q2.csv")
+# FIXME: drop the sentinel rows before release
+clean <- raw[raw$value >= 0, ]
+```
+
+The clean set has `r nrow(clean)` rows and `r ncol(clean)` columns, and the
+running total is `r sum(clean$value) # FIXME`.
+
+## Methods
+
+1. Fit the model with `lm()`.
+2. Check the residuals; XXX flag outliers by hand for now.
+3. Report $R^2$ and the fitted coefficients.
+
+> Estimates below the detection limit are shown as `<LOD`.
+
+```{r model, echo=FALSE, fig.cap="Fitted values against residuals."}
+fit <- lm(value ~ month, data = clean)
+plot(fit, which = 1) # XXX: swap for a ggplot version
+```
+
+## Results
+
+| Month | Value |
+|-------|-------|
+| May   | 12.4  |
+| June  | 13.1  |
+
+Table: Monthly means, TODO confirm the June figure.
+
+The slope is `r round(coef(fit)[2], 2)` per month, in line with earlier
+work [@smith2021].
+
+::: {.callout-note}
+These are preliminary figures; NOTE the caveats in the appendix.
+:::