feat(treesitter): bundle markdown parser and queries (#22481)

* bundle split Markdown parser from https://github.com/MDeiml/tree-sitter-markdown
* add queries from https://github.com/nvim-treesitter/nvim-treesitter/tree/main
* upstream `#trim!` and `#inject-language!` directives

Co-authored-by: dundargoc <gocdundar@gmail.com>

12 files changed, 325 insertions, 3 deletions

diff --git a/cmake.deps/cmake/BuildTreesitterParsers.cmake b/cmake.deps/cmake/BuildTreesitterParsers.cmake
index 56c8a5a7c6..89c1ec5cc9 100644
--- a/cmake.deps/cmake/BuildTreesitterParsers.cmake
+++ b/cmake.deps/cmake/BuildTreesitterParsers.cmake
@@ -37,3 +37,4 @@ endfunction()
 foreach(lang c lua vim vimdoc query)
   BuildTSParser(LANG ${lang})
 endforeach()
+BuildTSParser(LANG markdown CMAKE_FILE MarkdownParserCMakeLists.txt)
diff --git a/cmake.deps/cmake/MarkdownParserCMakeLists.txt b/cmake.deps/cmake/MarkdownParserCMakeLists.txt
new file mode 100644
index 0000000000..a356aaa391
--- /dev/null
+++ b/cmake.deps/cmake/MarkdownParserCMakeLists.txt
@@ -0,0 +1,28 @@
+cmake_minimum_required(VERSION 3.10)
+project(${PARSERLANG} C)
+
+set(CMAKE_C_STANDARD 99)
+
+add_library(markdown MODULE
+tree-sitter-markdown/src/parser.c
+tree-sitter-markdown/src/scanner.c)
+target_include_directories(markdown
+  PRIVATE
+  tree-sitter-markdown/src)
+
+add_library(markdown_inline MODULE
+tree-sitter-markdown-inline/src/parser.c
+tree-sitter-markdown-inline/src/scanner.c)
+target_include_directories(markdown_inline
+  PRIVATE
+  tree-sitter-markdown-inline/src)
+
+set_target_properties(
+  markdown markdown_inline
+  PROPERTIES
+    PREFIX ""
+)
+
+install(TARGETS markdown markdown_inline LIBRARY DESTINATION lib/nvim/parser)
+
+# vim: set ft=cmake:
diff --git a/cmake.deps/deps.txt b/cmake.deps/deps.txt
index 57b577dc33..fd0b4aa400 100644
--- a/cmake.deps/deps.txt
+++ b/cmake.deps/deps.txt
@@ -54,5 +54,7 @@ TREESITTER_VIMDOC_URL https://github.com/neovim/tree-sitter-vimdoc/archive/v2.0.
 TREESITTER_VIMDOC_SHA256 61e165df29778dc0c9277c2a7bc67447cc4e1bed36ca916a2f476dd25ce3260e
 TREESITTER_QUERY_URL https://github.com/nvim-treesitter/tree-sitter-query/archive/v0.1.0.tar.gz
 TREESITTER_QUERY_SHA256 e2b806f80e8bf1c4f4e5a96248393fe6622fc1fc6189d6896d269658f67f914c
+TREESITTER_MARKDOWN_URL https://github.com/MDeiml/tree-sitter-markdown/archive/936cc84289f6de83c263ae8e659fb342867ceb16.tar.gz
+TREESITTER_MARKDOWN_SHA256 4f2315930dc2c1bd42971a0b728cf4dafc57830c61f8abe3e2548cf230968713
 TREESITTER_URL https://github.com/tree-sitter/tree-sitter/archive/91e4d940169a0c0b024560632ef53c4f119117ca.tar.gz
 TREESITTER_SHA256 e15e335d127d38aaa73e727f3169df6015f43de1010d806e69b9e9222ad50fe1
diff --git a/runtime/doc/news.txt b/runtime/doc/news.txt
index 0a9b926b7f..5ee9cb7a14 100644
--- a/runtime/doc/news.txt
+++ b/runtime/doc/news.txt
@@ -101,6 +101,8 @@ The following new APIs and features were added.
 • Implemented LSP inlay hints: |vim.lsp.inlay_hint()|
   https://microsoft.github.io/language-server-protocol/specifications/lsp/3.17/specification/#textDocument_inlayHint
 
+• Bundled Markdown parser for treesitter highlighting and folding.
+
 ==============================================================================
 CHANGED FEATURES                                                 *news-changed*
 
@@ -120,7 +122,8 @@ The following changes to existing APIs or features add new behavior.
   • Automatic linting of treesitter query files (see |ft-query-plugin|).
     Can be disabled via: >lua
       vim.g.query_lint_on = {}
-<  • Enabled treesitter highlighting for treesitter query files.
+<
+  • Enabled treesitter highlighting for treesitter query files.
 
 • The `workspace/didChangeWatchedFiles` LSP client capability is now enabled
   by default.
diff --git a/runtime/doc/treesitter.txt b/runtime/doc/treesitter.txt
index 28ab3ecd8e..7539417197 100644
--- a/runtime/doc/treesitter.txt
+++ b/runtime/doc/treesitter.txt
@@ -292,7 +292,39 @@ The following directives are built in:
         Example: >query
             ((identifier) @constant (#offset! @constant 0 1 0 -1))
 <
+    `gsub!`                                          *treesitter-directive-gsub!*
+        Transforms the content of the node using a Lua pattern. This will set
+        a new `metadata[capture_id].text`.
 
+        Parameters: ~
+            {capture_id}
+            {pattern}
+
+        Example: >query
+            (#gsub! @_node ".*%.(.*)" "%1")
+<
+    `trim!`                                          *treesitter-directive-trim!*
+        Trim blank lines from the end of the node. This will set a new
+        `metadata[capture_id].range`.
+
+        Parameters: ~
+            {capture_id}
+
+        Example: >query
+            (#inject-language! @_lang)
+<
+    `inject-language!`                    *treesitter-directive-inject-language!*
+        Set the injection language from the node text, interpreted first as a
+        language name, then (if a parser is not found) a filetype. Custom
+        aliases can be added via |vim.treesitter.language.register()|. This
+        will set a new `metadata[capture_id]['injection.language']`.
+
+        Parameters: ~
+            {capture_id}
+
+        Example: >query
+            (#inject-language! @_lang)
+<
 Further directives can be added via |vim.treesitter.query.add_directive()|.
 Use |vim.treesitter.query.list_directives()| to list all available directives.
 
diff --git a/runtime/lua/vim/treesitter/query.lua b/runtime/lua/vim/treesitter/query.lua
index 73b561c777..7f90fa10e8 100644
--- a/runtime/lua/vim/treesitter/query.lua
+++ b/runtime/lua/vim/treesitter/query.lua
@@ -475,7 +475,6 @@ local directive_handlers = {
       metadata[capture_id].range = range
     end
   end,
-
   -- Transform the content of the node
   -- Example: (#gsub! @_node ".*%.(.*)" "%1")
   ['gsub!'] = function(match, _, bufnr, pred, metadata)
@@ -497,6 +496,65 @@ local directive_handlers = {
 
     metadata[id].text = text:gsub(pattern, replacement)
   end,
+  -- Trim blank lines from end of the node
+  -- Example: (#trim! @fold)
+  -- TODO(clason): generalize to arbitrary whitespace removal
+  ['trim!'] = function(match, _, bufnr, pred, metadata)
+    local node = match[pred[2]]
+    if not node then
+      return
+    end
+
+    local start_row, start_col, end_row, end_col = node:range()
+
+    -- Don't trim if region ends in middle of a line
+    if end_col ~= 0 then
+      return
+    end
+
+    while true do
+      -- As we only care when end_col == 0, always inspect one line above end_row.
+      local end_line = vim.api.nvim_buf_get_lines(bufnr, end_row - 1, end_row, true)[1]
+
+      if end_line ~= '' then
+        break
+      end
+
+      end_row = end_row - 1
+    end
+
+    -- If this produces an invalid range, we just skip it.
+    if start_row < end_row or (start_row == end_row and start_col <= end_col) then
+      metadata.range = { start_row, start_col, end_row, end_col }
+    end
+  end,
+  -- Set injection language from node text, interpreted first as language and then as filetype
+  -- Example: (#inject-language! @_lang)
+  ['inject-language!'] = function(match, _, bufnr, pred, metadata)
+    local id = pred[2]
+    local node = match[id]
+    if not node then
+      return
+    end
+
+    -- TODO(clason): replace by refactored `ts.has_parser` API
+    local has_parser = function(lang)
+      return vim._ts_has_language(lang)
+        or #vim.api.nvim_get_runtime_file('parser/' .. lang .. '.*', false) > 0
+    end
+
+    local alias = vim.treesitter.get_node_text(node, bufnr, { metadata = metadata[id] })
+    if not alias then
+      return
+    elseif has_parser(alias) then
+      metadata['injection.language'] = alias
+    else
+      local lang = vim.treesitter.language.get_lang(alias)
+      if lang and has_parser(lang) then
+        metadata['injection.language'] = lang
+      end
+    end
+  end,
 }
 
 --- Adds a new predicate to be used in queries
diff --git a/runtime/queries/markdown/folds.scm b/runtime/queries/markdown/folds.scm
new file mode 100644
index 0000000000..5900f7ffbe
--- /dev/null
+++ b/runtime/queries/markdown/folds.scm
@@ -0,0 +1,9 @@
+(
+  [
+    (fenced_code_block)
+    (indented_code_block)
+    (list)
+    (section)
+  ] @fold
+  (#trim! @fold)
+)
diff --git a/runtime/queries/markdown/highlights.scm b/runtime/queries/markdown/highlights.scm
new file mode 100644
index 0000000000..e78d233cc6
--- /dev/null
+++ b/runtime/queries/markdown/highlights.scm
@@ -0,0 +1,63 @@
+;From MDeiml/tree-sitter-markdown & Helix
+(setext_heading (paragraph) @text.title.1 (setext_h1_underline) @text.title.1.marker)
+(setext_heading (paragraph) @text.title.2 (setext_h2_underline) @text.title.2.marker)
+
+(atx_heading (atx_h1_marker) @text.title.1.marker (inline) @text.title.1)
+(atx_heading (atx_h2_marker) @text.title.2.marker (inline) @text.title.2)
+(atx_heading (atx_h3_marker) @text.title.3.marker (inline) @text.title.3)
+(atx_heading (atx_h4_marker) @text.title.4.marker (inline) @text.title.4)
+(atx_heading (atx_h5_marker) @text.title.5.marker (inline) @text.title.5)
+(atx_heading (atx_h6_marker) @text.title.6.marker (inline) @text.title.6)
+
+(link_title) @text.literal
+(indented_code_block) @text.literal.block
+((fenced_code_block) @text.literal.block (#set! "priority" 90))
+
+(info_string) @label
+
+(pipe_table_header (pipe_table_cell) @text.title)
+
+(pipe_table_header "|" @punctuation.special)
+(pipe_table_row "|" @punctuation.special)
+(pipe_table_delimiter_row "|" @punctuation.special)
+(pipe_table_delimiter_cell) @punctuation.special
+
+[
+  (fenced_code_block_delimiter)
+] @punctuation.delimiter
+
+(code_fence_content) @none
+
+[
+  (link_destination)
+] @text.uri
+
+[
+  (link_label)
+] @text.reference
+
+[
+  (list_marker_plus)
+  (list_marker_minus)
+  (list_marker_star)
+  (list_marker_dot)
+  (list_marker_parenthesis)
+  (thematic_break)
+] @punctuation.special
+
+
+(task_list_marker_unchecked) @text.todo.unchecked
+(task_list_marker_checked) @text.todo.checked
+
+(block_quote) @text.quote
+
+[
+  (block_continuation)
+  (block_quote_marker)
+] @punctuation.special
+
+[
+  (backslash_escape)
+] @string.escape
+
+(inline) @spell
diff --git a/runtime/queries/markdown/injections.scm b/runtime/queries/markdown/injections.scm
new file mode 100644
index 0000000000..0bead6f4ac
--- /dev/null
+++ b/runtime/queries/markdown/injections.scm
@@ -0,0 +1,26 @@
+(fenced_code_block
+  (info_string
+    (language) @_lang)
+  (code_fence_content) @injection.content
+  (#inject-language! @_lang))
+
+((html_block) @injection.content 
+ (#set! injection.language "html")
+ (#set! injection.combined)
+ (#set! injection.include-children))
+
+((minus_metadata) @injection.content 
+ (#set! injection.language "yaml") 
+ (#offset! @injection.content 1 0 -1 0)
+ (#set! injection.include-children))
+
+((plus_metadata) @injection.content 
+ (#set! injection.language "toml") 
+ (#offset! @injection.content 1 0 -1 0)
+ (#set! injection.include-children))
+
+([
+  (inline)
+  (pipe_table_cell)
+ ] @injection.content
+ (#set! injection.language "markdown_inline"))
diff --git a/runtime/queries/markdown_inline/highlights.scm b/runtime/queries/markdown_inline/highlights.scm
new file mode 100644
index 0000000000..a70e34bb87
--- /dev/null
+++ b/runtime/queries/markdown_inline/highlights.scm
@@ -0,0 +1,92 @@
+;; From MDeiml/tree-sitter-markdown
+[
+  (code_span)
+  (link_title)
+] @text.literal @nospell
+
+[
+  (emphasis_delimiter)
+  (code_span_delimiter)
+] @punctuation.delimiter
+
+(emphasis) @text.emphasis
+
+(strong_emphasis) @text.strong
+
+(strikethrough) @text.strike
+
+[
+  (link_destination)
+  (uri_autolink)
+] @text.uri @nospell
+
+[
+  (link_label)
+  (link_text)
+  (image_description)
+] @text.reference
+
+[
+  (backslash_escape)
+  (hard_line_break)
+] @string.escape
+
+(image "!" @punctuation.special)
+(image ["[" "]" "(" ")"] @punctuation.bracket)
+(inline_link ["[" "]" "(" ")"] @punctuation.bracket)
+(shortcut_link ["[" "]"] @punctuation.bracket)
+
+; Conceal codeblock and text style markers
+([
+  (code_span_delimiter)
+  (emphasis_delimiter)
+] @conceal
+(#set! conceal ""))
+
+; Conceal inline links
+(inline_link
+  [
+    "["
+    "]"
+    "("
+    (link_destination)
+    ")"
+  ] @conceal
+  (#set! conceal ""))
+
+; Conceal image links
+(image
+  [
+    "!"
+    "["
+    "]"
+    "("
+    (link_destination)
+    ")"
+  ] @conceal
+  (#set! conceal ""))
+
+; Conceal full reference links
+(full_reference_link
+  [
+    "["
+    "]"
+    (link_label)
+  ] @conceal
+  (#set! conceal ""))
+
+; Conceal collapsed reference links
+(collapsed_reference_link
+  [
+    "["
+    "]"
+  ] @conceal
+  (#set! conceal ""))
+
+; Conceal shortcut links
+(shortcut_link
+  [
+    "["
+    "]"
+  ] @conceal
+  (#set! conceal ""))
diff --git a/runtime/queries/markdown_inline/injections.scm b/runtime/queries/markdown_inline/injections.scm
new file mode 100644
index 0000000000..f7aa19caff
--- /dev/null
+++ b/runtime/queries/markdown_inline/injections.scm
@@ -0,0 +1,8 @@
+((html_tag) @injection.content
+ (#set! injection.language "html")
+ (#set! injection.combined)
+ (#set! injection.include-children))
+
+((latex_block) @injection.content
+ (#set! injection.language "latex")
+ (#set! injection.include-children))
diff --git a/test/functional/treesitter/parser_spec.lua b/test/functional/treesitter/parser_spec.lua
index d2f40bfe00..7cfe5b69de 100644
--- a/test/functional/treesitter/parser_spec.lua
+++ b/test/functional/treesitter/parser_spec.lua
@@ -783,7 +783,7 @@ int x = INT_MAX;
         return list
         ]]
 
-        eq({ 'gsub!', 'offset!', 'set!' }, res_list)
+        eq({ 'gsub!', 'inject-language!', 'offset!', 'set!', 'trim!' }, res_list)
       end)
     end)
   end)