refactor(tree-sitter): centralize grammar config & hacks

Easier to track and maintain.

modules/lang/cc/config.el

@@ -30,10 +30,8 @@ This is ignored by ccls.")
   :hook (c-mode-common . rainbow-delimiters-mode)
   :init
   (when (modulep! +tree-sitter)
-    (set-tree-sitter! 'c-mode 'c-ts-mode
+    (set-tree-sitter! 'c-mode 'c-ts-mode 'c)
-      '((c :url "https://github.com/tree-sitter/tree-sitter-c")))
+    (set-tree-sitter! 'c++-mode 'c++-ts-mode 'cpp))
-    (set-tree-sitter! 'c++-mode 'c++-ts-mode
-      '((cpp :url "https://github.com/tree-sitter/tree-sitter-cpp"))))
 
   :config
   (set-docsets! '(c-mode c-ts-mode) "C")
@@ -116,8 +114,7 @@ This is ignored by ccls.")
   :init
   (when (and (modulep! +tree-sitter)
              (boundp 'cmake-ts-mode)) ; 29+ only
-    (set-tree-sitter! 'cmake-mode 'cmake-ts-mode
+    (set-tree-sitter! 'cmake-mode 'cmake-ts-mode 'cmake))
-      '((cmake :url "https://github.com/uyha/tree-sitter-cmake"))))
   :config
   (set-docsets! '(cmake-mode cmake-ts-mode) "CMake")
   (set-popup-rule! "^\\*CMake Help\\*" :size 0.4 :ttl t)
@@ -160,12 +157,7 @@ This is ignored by ccls.")
     '((cuda :url "https://github.com/tree-sitter-grammars/tree-sitter-cuda")))
   :config
   (when (modulep! +lsp)
-    (add-hook 'cuda-ts-mode-local-vars-hook #'lsp! 'append))
+    (add-hook 'cuda-ts-mode-local-vars-hook #'lsp! 'append)))
-  ;; HACK: Remove redundant entries so we can rely solely on
-  ;;   `major-mode-remap-defaults' et co.
-  (rassq-delete-all 'cuda-ts-mode auto-mode-alist)
-  (cl-callf2 delete '(cuda "https://github.com/tree-sitter-grammars/tree-sitter-cuda" nil nil nil nil)
-             treesit-language-source-alist))
 
 
 (use-package! demangle-mode

modules/lang/clojure/config.el

@@ -62,7 +62,6 @@
   ;; HACK: Rely on `major-mode-remap-defaults' instead (upstream also doesn't
   ;;   check if the grammars are ready before adding these entries, which will
   ;;   bork clojure buffers.
-  (cl-callf2 rassq-delete-all 'clojure-ts-mode auto-mode-alist)
   (cl-callf2 rassq-delete-all 'clojure-ts-clojurescript-mode auto-mode-alist)
   (cl-callf2 rassq-delete-all 'clojure-ts-clojurec-mode auto-mode-alist)
   (cl-callf2 rassq-delete-all 'clojure-ts-clojuredart-mode auto-mode-alist)

modules/lang/csharp/config.el

@@ -2,6 +2,9 @@
 
 (use-package! csharp-mode
   :hook (csharp-mode . rainbow-delimiters-mode)
+  :init
+  (when (modulep! +tree-sitter)
+    (set-tree-sitter! 'csharp-mode 'csharp-ts-mode 'c-sharp))
   :config
   (set-formatter! 'csharpier '("csharpier" "format" "--write-stdout")
     :modes '(csharp-mode csharp-ts-mode))
@@ -38,26 +41,12 @@
     (add-hook 'csharp-mode-local-vars-hook #'lsp! 'append)
     (add-hook 'csharp-ts-mode-local-vars-hook #'lsp! 'append))
 
-  (when (and (modulep! +tree-sitter)
-             (fboundp 'csharp-ts-mode)) ; 29.1+ only
-    (set-tree-sitter! 'csharp-mode 'csharp-ts-mode
-      '((c-sharp :url "https://github.com/tree-sitter/tree-sitter-c-sharp"
-                 :rev "v0.23.1"))))
-
   (defadvice! +csharp-disable-clear-string-fences-a (fn &rest args)
     "This turns off `c-clear-string-fences' for `csharp-mode'. When
 on for `csharp-mode' font lock breaks after an interpolated string
 or terminating simple string."
     :around #'csharp-disable-clear-string-fences
     (unless (eq major-mode 'csharp-mode)
-      (apply fn args)))
-
-  ;; HACK: `csharp-ts-mode' changes `auto-mode-alist' every time the mode is
-  ;;   activated, which runs the risk of overwriting user (or Doom) entries.
-  ;; REVIEW: Should be addressed upstream.
-  (defadvice! +csharp--undo-ts-side-effects-a (fn &rest args)
-    :around #'csharp-ts-mode
-    (let (auto-mode-alist)
       (apply fn args))))
 
 

modules/lang/elixir/config.el

@@ -59,16 +59,8 @@
   :when (fboundp 'elixir-ts-mode) ; 30.1+ only
   :defer t
   :init
-  (set-tree-sitter! 'elixir-mode 'elixir-ts-mode
+  (set-tree-sitter! 'elixir-mode 'elixir-ts-mode '(elixir heex))
-    '((elixir :url "https://github.com/elixir-lang/tree-sitter-elixir"
-              :rev "v0.3.3")
-      (heex :url "https://github.com/phoenixframework/tree-sitter-heex"
-            :rev "v0.7.0")))
   :config
-  ;; HACK: Rely on `major-mode-remap-defaults' (and elixir-mode's autoloaded
-  ;;   auto-mode-alist entries).
-  (cl-callf2 rassq-delete-all 'elixir-ts-mode auto-mode-alist)
-
   (+elixir-common-config 'elixir-ts-mode))
 
 

modules/lang/go/config.el

@@ -53,10 +53,7 @@
   :when (fboundp 'go-ts-mode) ; 31.1+ only
   :defer t
   :init
-  (set-tree-sitter! 'go-mode 'go-ts-mode
+  (set-tree-sitter! 'go-mode 'go-ts-mode '(go gomod gowork))
-    '((go :url "https://github.com/tree-sitter/tree-sitter-go" :ref "v0.23.4")
-      (gomod :url "https://github.com/camdencheek/tree-sitter-go-mod" :ref "v1.1.0")
-      (gowork :url "https://github.com/omertuc/tree-sitter-go-work")))
   :config
   (+go-common-config 'go-ts-mode))
 

modules/lang/janet/config.el

@@ -30,9 +30,4 @@
   :init
   (set-tree-sitter! 'janet-mode 'janet-ts-mode
     `(janet-simple :url "https://github.com/sogaiu/tree-sitter-janet-simple"
-                   :cc ,(if (featurep :system 'windows) "gcc.exe")))
+                   :cc ,(if (featurep :system 'windows) "gcc.exe"))))
-  :config
-  ;; HACK: These entries are inserted twice by this package, so remove them so
-  ;;   this module can be the single source of truth.
-  (cl-callf2 rassq-delete-all 'janet-ts-mode auto-mode-alist)
-  (cl-callf2 rassq-delete-all 'janet-ts-mode interpreter-mode-alist))

modules/lang/json/config.el

@@ -25,9 +25,7 @@
   :when (fboundp 'json-ts-mode) ; 29.1+ only
   :defer t
   :init
-  (set-tree-sitter! 'json-mode 'json-ts-mode
+  (set-tree-sitter! 'json-mode 'json-ts-mode 'json)
-    '((json :url "https://github.com/tree-sitter/tree-sitter-json"
-            :rev "v0.24.8")))
   :config
   ;; HACK: Rely on `major-mode-remap-defaults'.
   (cl-callf2 assq-delete-all 'json-ts-mode auto-mode-alist)

modules/lang/julia/config.el

@@ -48,9 +48,7 @@
   :when (modulep! +tree-sitter)
   :defer t
   :init
-  (set-tree-sitter! 'julia-mode 'julia-ts-mode
+  (set-tree-sitter! 'julia-mode 'julia-ts-mode 'julia)
-    '((julia :url "https://github.com/tree-sitter/tree-sitter-julia"
-             :rev "v0.23.1")))
   :config
   (when (modulep! +lsp)
     (add-hook 'julia-ts-mode-local-vars-hook #'lsp! 'append)))

modules/lang/lua/config.el

@@ -30,9 +30,7 @@
   :when (fboundp 'lua-ts-mode) ; 30.1+ only
   :defer t
   :init
-  (set-tree-sitter! 'lua-mode 'lua-ts-mode
+  (set-tree-sitter! 'lua-mode 'lua-ts-mode 'lua)
-    '((lua :url "https://github.com/tree-sitter-grammars/tree-sitter-lua"
-           :rev "v0.3.0")))
   :config
   (set-lookup-handlers! 'lua-ts-mode :documentation 'lua-search-documentation)
   (set-electric! 'lua-ts-mode :words '("else" "end"))

modules/lang/markdown/config.el

@@ -131,14 +131,7 @@ capture, the end position, and the output buffer.")
   :when (fboundp 'markdown-ts-mode)
   :defer t
   :init
-  (set-tree-sitter! 'markdown-mode 'markdown-ts-mode
+  (set-tree-sitter! 'markdown-mode 'markdown-ts-mode '(markdown markdown-inline))
-    '((markdown :url "https://github.com/tree-sitter-grammars/tree-sitter-markdown"
-                :rev "v0.4.1"
-                :source-dir "tree-sitter-markdown-inline/src")
-      (markdown-inline :url "https://github.com/tree-sitter-grammars/tree-sitter-markdown"
-                       :rev "v0.4.1"
-                       :source-dir "tree-sitter-markdown-inline/src")))
-
   :config
   (cl-callf2 delete '("\\.md\\'" . markdown-ts-mode) auto-mode-alist))
 

modules/lang/php/config.el

@@ -83,16 +83,8 @@
   :when (fboundp 'php-ts-mode) ; 30.1+ only
   :defer t
   :init
-  (set-tree-sitter! 'php-mode 'php-ts-mode
+  (set-tree-sitter! 'php-mode 'php-ts-mode '(php phpdoc))
-    '((php :url "https://github.com/tree-sitter/tree-sitter-php"
-       :rev "v0.23.11"
-       :source-dir "php/src")
-      (phpdoc :url "https://github.com/claytonrcarter/tree-sitter-phpdoc")))
   :config
-  ;; HACK: Rely on `major-mode-remap-defaults'.
-  (cl-callf2 rassq-delete-all 'php-ts-mode auto-mode-alist)
-  (cl-callf2 rassq-delete-all 'php-ts-mode interpreter-mode-alist)
-
   (+php-common-config 'php-ts-mode))
 
 

modules/lang/python/config.el

@@ -20,8 +20,7 @@
         python-indent-guess-indent-offset-verbose nil)
 
   (when (modulep! +tree-sitter)
-    (set-tree-sitter! 'python-mode 'python-ts-mode
+    (set-tree-sitter! 'python-mode 'python-ts-mode 'python))
-      '((python :url "https://github.com/tree-sitter/tree-sitter-python"))))
 
   :config
   ;; HACK: `python-base-mode' (and `python-ts-mode') don't exist on pre-29
@@ -98,16 +97,7 @@
 
   ;; HACK: `python-mode' doesn't update `tab-width' to reflect
   ;;   `python-indent-offset', causing issues anywhere `tab-width' is respected.
-  (setq-hook! '(python-mode-hook python-ts-mode-hook) tab-width python-indent-offset)
+  (setq-hook! '(python-mode-hook python-ts-mode-hook) tab-width python-indent-offset))
-
-  ;; HACK: `python-ts-mode' changes `auto-mode-alist' and
-  ;;   `interpreter-mode-alist' every time the mode is activated, which runs the
-  ;;   risk of overwriting user (or Doom) entries.
-  ;; REVIEW: Should be addressed upstream.
-  (defadvice! +python--undo-ts-side-effects-a (fn &rest args)
-    :around #'python-ts-mode
-    (let (auto-mode-alist interpreter-mode-alist)
-      (apply fn args))))
 
 
 (use-package! pyimport

modules/lang/ruby/config.el

@@ -41,8 +41,7 @@
   :when (fboundp 'ruby-ts-mode) ; 29.1+ only
   :defer t
   :init
-  (set-tree-sitter! 'ruby-mode 'ruby-ts-mode
+  (set-tree-sitter! 'ruby-mode 'ruby-ts-mode 'ruby)
-    '((ruby :url "https://github.com/tree-sitter/tree-sitter-ruby")))
   :config
   (set-electric! 'ruby-ts-mode :words '("else" "end" "elsif"))
   (set-repl-handler! 'ruby-ts-mode #'inf-ruby)

modules/lang/yaml/config.el

@@ -12,12 +12,7 @@
   :when (modulep! +tree-sitter)
   :when (fboundp 'yaml-ts-mode) ; 29.1+ only
   :init
-  (set-tree-sitter! 'yaml-mode 'yaml-ts-mode
+  (set-tree-sitter! 'yaml-mode 'yaml-ts-mode 'yaml)
-    '((yaml :url "https://github.com/tree-sitter-grammars/tree-sitter-yaml"
-            :rev "v0.7.0")))
   :config
-  ;; HACK: Rely on `major-mode-remap-defaults'.
-  (cl-callf2 rassq-delete-all 'yaml-ts-mode auto-mode-alist)
-
   (when (modulep! +lsp)
     (add-hook 'yaml-ts-mode-local-vars-hook #'lsp! 'append)))

modules/lang/zig/config.el

@@ -47,7 +47,4 @@
     '((zig :url "https://github.com/tree-sitter/zig-tree-sitter"
            :rev "v0.25.0")))
   :config
-  ;; HACK: Rely on `major-mode-remap-defaults'
-  (cl-callf2 rassq-delete-all 'zig-ts-mode auto-mode-alist)
-
   (+zig-common-config 'zig-ts-mode))

modules/tools/docker/config.el

@@ -13,9 +13,4 @@
   :when (fboundp 'dockerfile-ts-mode) ; 29.1+ only
   :defer t
   :init
-  (set-tree-sitter! 'dockerfile-mode 'dockerfile-ts-mode
+  (set-tree-sitter! 'dockerfile-mode 'dockerfile-ts-mode 'dockerfile))
-    '((dockerfile :url "https://github.com/camdencheek/tree-sitter-dockerfile"
-                  :rev "v0.2.0")))
-  :config
-  ;; HACK: Rely on `major-mode-remap-defaults' instead
-  (cl-callf2 rassq-delete-all 'dockerfile-ts-mode auto-mode-alist))

modules/tools/tree-sitter/autoload/tree-sitter.el

@@ -15,10 +15,11 @@ MODE and TS-MODE are major mode symbols. If RECIPES is provided, fall back to
 MODE if RECIPES don't pass `treesit-ready-p' when activating TS-MODE. Use this
 for ts modes that error out instead of failing gracefully.
 
-RECIPES are an alist of plists with the format (LANG &key URL REV SOURCE-DIR CC
+RECIPES is a symbol (a grammar language name), list thereof, or alist of plists
-CPP COMMIT), which will be transformed into entries for
+with the format (LANG &key URL REV SOURCE-DIR CC CPP COMMIT). If an alist of
-`treesit-language-source-alist' (which descrie what each of these keys mean).
+plists, it will be transformed into entries for `treesit-language-source-alist'
-Note that COMMIT is only available in Emacs >=31."
+(which describe what each of these keys mean). Note that COMMIT is ignored
+pre-Emacs 31."
   (declare (indent 2))
   (cl-check-type mode symbol)
   (cl-check-type ts-mode symbol)
@@ -41,6 +42,11 @@ Note that COMMIT is only available in Emacs >=31."
                          (fboundp ts-mode)
                          (or (eq treesit-enabled-modes t)
                              (memq ts-mode treesit-enabled-modes))
+                         ;; Lazily load autoload so
+                         ;; `treesit-language-source-alist' is initialized.
+                         (let ((fn (symbol-function ts-mode)))
+                           (or (not (autoloadp fn))
+                               (autoload-do-load fn)))
                          ;; Only prompt once, and log other times.
                          (cl-every (if ensured?
                                        (doom-rpartial #'treesit-ready-p 'message)
@@ -54,16 +60,16 @@ Note that COMMIT is only available in Emacs >=31."
                     m))))))))
   (with-eval-after-load 'treesit
     (dolist (recipe recipes)
-      (cl-destructuring-bind (name &key url rev source-dir cc cpp commit)
+      (when (cdr (setq recipe (ensure-list recipe)))
-          (ensure-list recipe)
+        (cl-destructuring-bind (name &key url rev source-dir cc cpp commit) recipe
-        (setf (alist-get name treesit-language-source-alist)
+          (setf (alist-get name treesit-language-source-alist)
-              (append (list url rev source-dir cc cpp)
+                (append (list url rev source-dir cc cpp)
-                      ;; COMPAT: 31.1 introduced a COMMIT recipe argument. On
+                        ;; COMPAT: 31.1 introduced a COMMIT recipe argument. On
-                      ;;   <=30.x, extra arguments will trigger an arity error
+                        ;;   <=30.x, extra arguments will trigger an arity error
-                      ;;   when installing grammars.
+                        ;;   when installing grammars.
-                      (if (eq (cdr (func-arity 'treesit--install-language-grammar-1))
+                        (if (eq (cdr (func-arity 'treesit--install-language-grammar-1))
-                              'many)
+                                'many)
-                          (list commit))))))))
+                            (list commit)))))))))
 
 ;; ;; HACK: Remove and refactor when `use-package' eager macro expansion is solved or `use-package!' is removed
 ;; ;;;###autoload
@@ -81,6 +87,11 @@ Note that COMMIT is only available in Emacs >=31."
 ;;                 (evil-textobj-tree-sitter-goto-textobj group previous end query)))
 ;;     sym))
 
+;;;###autoload
+(defun +tree-sitter-ts-mode-inhibit-side-effects-a (fn &rest args)
+  "Suppress changes to `auto-mode-alist' and `interpreter-mode-alist'."
+  (let (auto-mode-alist interpreter-mode-alist)
+    (apply fn args)))
 
 ;;; TODO: Backwards compatibility
 

modules/tools/tree-sitter/config.el

@@ -10,38 +10,42 @@
   :preface
   (setq treesit-enabled-modes t)
 
-  ;; HACK: These *-ts-mode-maybe functions all treat `treesit-enabled-modes'
+  ;; HACK: The *-ts-mode major modes are inconsistent about how they treat
-  ;;   strangely in the event the language's grammar is unavailable. Plus, they
+  ;;   missing language grammars (some error out, some respect
-  ;;   add yet-another-layer of complexity for users to be cognicent of. Get rid
+  ;;   `treesit-auto-install-grammar', some fall back to `fundamental-mode').
-  ;;   of them.
+  ;;   I'd like to address this poor UX using `major-mode-remap-alist' entries
-  ;; REVIEW: Handle this during the 'doom sync' process instead.
+  ;;   created by `set-tree-sitter!' (which will fall back to the non-treesit
-  (setq auto-mode-alist
+  ;;   modes), but most *-ts-mode's clobber `auto-mode-alist' and/or
-        (save-match-data
+  ;;   `interpreter-mode-alist' each time the major mode is activated, so those
-          (cl-loop for (src . fn) in auto-mode-alist
+  ;;   must be undone too so they don't overwrite user config.
-                   unless (and (functionp fn)
+  ;; TODO: Handle this during the 'doom sync' process instead.
-                               (string-match "-ts-mode-maybe$" (symbol-name fn)))
+  (save-match-data
-                   collect (cons src fn))))
+    (dolist (sym '(auto-mode-alist interpreter-mode-alist))
+      (set
+       sym (cl-loop for (src . fn) in (symbol-value sym)
+                    unless (and (functionp fn)
+                                (string-match "-ts-mode\\(?:-maybe\\)?$" (symbol-name fn)))
+                    collect (cons src fn)))))
+
+  ;; HACK: These *-ts-modes change `auto-mode-alist' and/or
+  ;;   `interpreter-mode-alist' every time they are activated, running the risk
+  ;;   of overwriting user (or Doom) config.
+  ;; REVIEW: Should be addressed upstream.
+  (dolist (mode '(csharp-ts-mode
+                  python-ts-mode))
+    (advice-add mode :around #'+tree-sitter-ts-mode-inhibit-side-effects-a))
 
   :config
   ;; HACK: The implementation of `treesit-enabled-modes's setter and
-  ;;   `treesit-major-mode-remap-alist' are intrusively opinionated, so I
+  ;;   `treesit-major-mode-remap-alist' is intrusively opinionated, so disable
-  ;;   disable it altogether as to not unexpectedly modify
+  ;;   it ato avoid untimely (and overriding) modifications of
-  ;;   `major-mode-remap-alist' at runtime. What's more, there's no guarantee
+  ;;   `major-mode-remap-alist' at runtime. What's more, this was only
-  ;;   this will be populated correctly unless the user is on a particular
+  ;;   introduced in 31, so ignoring them is more consistent for pre-31 users.
-  ;;   commit of Emacs 31 or newer. Best we simply ignore it.
+  (when major-mode-remap-alist
-  (dolist (m treesit-major-mode-remap-alist)
+    (dolist (m treesit-major-mode-remap-alist)
-    (setq major-mode-remap-alist (delete m major-mode-remap-alist)))
+      (setq major-mode-remap-alist (delete m major-mode-remap-alist))))
   (setq treesit-major-mode-remap-alist nil)
 
-  ;; HACK: treesit lacks any way to dictate where to install grammars.
-  (add-to-list 'treesit-extra-load-path (concat doom-profile-data-dir "tree-sitter"))
-  (defadvice! +tree-sitter--install-grammar-to-local-dir-a (fn &rest args)
-    "Write grammars to `doom-profile-data-dir'."
-    :around #'treesit-install-language-grammar
-    :around #'treesit--build-grammar
-    (let ((user-emacs-directory doom-profile-data-dir))
-      (apply fn args)))
-
   ;; HACK: Some *-ts-mode packages modify `major-mode-remap-defaults'
   ;;   inconsistently. Playing whack-a-mole to undo those changes is more hassle
   ;;   then simply ignoring them (by overriding `major-mode-remap-defaults' for
@@ -56,6 +60,15 @@
              major-mode-remap-defaults)))
       (funcall fn mode)))
 
+  ;; HACK: Keep $EMACSDIR clean by installing grammars to the active profile.
+  (add-to-list 'treesit-extra-load-path (concat doom-profile-data-dir "tree-sitter"))
+  (defadvice! +tree-sitter--install-grammar-to-local-dir-a (fn &rest args)
+    "Write grammars to `doom-profile-data-dir'."
+    :around #'treesit-install-language-grammar
+    :around #'treesit--build-grammar
+    (let ((user-emacs-directory doom-profile-data-dir))
+      (apply fn args)))
+
   ;; TODO: Move most of these out to modules
   (dolist (map '((awk "https://github.com/Beaglefoot/tree-sitter-awk" nil nil nil nil)
                  (bibtex "https://github.com/latex-lsp/tree-sitter-bibtex" nil nil nil nil)