feat(markdown): configured aliases for fenced code block languages (#4659)

* feat(markdown): configured aliases for languages

* refactor: use vim.treesitter.match

* refactor: rename local vars

* fix: query syntax

lua/nvim-treesitter/query_predicates.lua

@@ -1,5 +1,25 @@
 local query = require "vim.treesitter.query"
 
+local html_script_type_languages = {
+  ["importmap"] = "json",
+  ["module"] = "javascript",
+  ["application/ecmascript"] = "javascript",
+  ["text/ecmascript"] = "javascript",
+}
+
+local non_filetype_match_injection_language_aliases = {
+  ex = "elixir",
+  pl = "perl",
+  sh = "bash",
+  uxn = "uxntal",
+  ts = "typescript",
+}
+
+local function get_parser_from_markdown_info_string(injection_alias)
+  local match = vim.filetype.match { filename = "a." .. injection_alias }
+  return match or non_filetype_match_injection_language_aliases[injection_alias] or injection_alias
+end
+
 local function error(str)
   vim.api.nvim_err_writeln(str)
 end
@@ -119,19 +139,17 @@ query.add_predicate("has-type?", function(match, _pattern, _bufnr, pred)
   return vim.tbl_contains(types, node:type())
 end)
 
-local html_script_type_languages = {
+---@param match (TSNode|nil)[]
-  ["importmap"] = "json",
+---@param _ string
-  ["module"] = "javascript",
+---@param bufnr integer
-  ["application/ecmascript"] = "javascript",
+---@param pred string[]
-  ["text/ecmascript"] = "javascript",
-}
-
----@param match string
----@param metadata table
 ---@return boolean|nil
-query.add_directive("set-lang-from-mimetype!", function(match, pattern, bufnr, predicate, metadata)
+query.add_directive("set-lang-from-mimetype!", function(match, _, bufnr, pred, metadata)
-  local capture_id = predicate[2]
+  local capture_id = pred[2]
   local node = match[capture_id]
+  if not node then
+    return
+  end
   local type_attr_value = vim.treesitter.get_node_text(node, bufnr)
   local configured = html_script_type_languages[type_attr_value]
   if configured then
@@ -142,6 +160,21 @@ query.add_directive("set-lang-from-mimetype!", function(match, pattern, bufnr, p
   end
 end)
 
+---@param match (TSNode|nil)[]
+---@param _ string
+---@param bufnr integer
+---@param pred string[]
+---@return boolean|nil
+query.add_directive("set-lang-from-info-string!", function(match, _, bufnr, pred, metadata)
+  local capture_id = pred[2]
+  local node = match[capture_id]
+  if not node then
+    return
+  end
+  local injection_alias = vim.treesitter.get_node_text(node, bufnr)
+  metadata.language = get_parser_from_markdown_info_string(injection_alias)
+end)
+
 -- Just avoid some annoying warnings for this directive
 query.add_directive("make-range!", function() end)
 

queries/markdown/injections.scm

@@ -1,8 +1,11 @@
 (fenced_code_block
   (info_string
-    (language) @language)
+    (language) @_lang)
-  (#not-match? @language "elm")
+  (#not-match? @_lang "elm") ; prevent segfault when using elm parser
-  (code_fence_content) @content (#exclude_children! @content))
+  (code_fence_content)
+    @content
+    (#set-lang-from-info-string! @_lang)
+    (#exclude_children! @content))
 
 ((html_block) @html)