bug#74612: 30.0.90; [PATCH] Allow passing nil to treesit-node-match-p

bug#74612: 30.0.90; [PATCH] Allow passing nil to treesit-node-match-p

[Yuan Fu]

[-- Attachment #1: Type: text/plain, Size: 164 bytes --]

Eli, are you ok with this patch on emacs-30? Almost all tree-sitter node functions allows user to pass nil for NODE, so treesit-node-match-p should too.

Yuan


[-- Attachment #2: treesit-node-match-p.patch --]
[-- Type: application/octet-stream, Size: 1301 bytes --]

From 9ab20d03e30e46e2b8502877de68eec0b511d3da Mon Sep 17 00:00:00 2001
From: Yuan Fu <casouri@gmail.com>
Date: Fri, 29 Nov 2024 16:33:28 -0800
Subject: [PATCH] Allow passing nil to treesit-node-match-p

* src/treesit.c (Ftreesit_node_match_p): Return nil if NODE is nil.
---
 src/treesit.c | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)

diff --git a/src/treesit.c b/src/treesit.c
index 4031d80f7c9..cda6d4af2ee 100644
--- a/src/treesit.c
+++ b/src/treesit.c
@@ -4017,7 +4017,8 @@ DEFUN ("treesit-node-match-p",
 `treesit-thing-settings', or a predicate, like regexp matching node
 type, etc.  See `treesit-thing-settings' for more details.
 
-Return non-nil if NODE matches PREDICATE, nil otherwise.
+Return non-nil if NODE matches PREDICATE, nil otherwise.  If NODE is
+nil, return nil.
 
 Signals `treesit-invalid-predicate' if there's no definition of THING
 in `treesit-thing-settings', or if PREDICATE is malformed.  If
@@ -4025,6 +4026,8 @@ DEFUN ("treesit-node-match-p",
 definition, but still signal for malformed PREDICATE.  */)
   (Lisp_Object node, Lisp_Object predicate, Lisp_Object ignore_missing)
 {
+  if (NILP (node)) return Qnil;
+
   CHECK_TS_NODE (node);
 
   Lisp_Object parser = XTS_NODE (node)->parser;
-- 
2.39.5 (Apple Git-151)

bug#74612: 30.0.90; [PATCH] Allow passing nil to treesit-node-match-p

[Eli Zaretskii]

> From: Yuan Fu <casouri@gmail.com>
> Date: Fri, 29 Nov 2024 16:38:22 -0800
> 
> Eli, are you ok with this patch on emacs-30? Almost all tree-sitter node functions allows user to pass nil for NODE, so treesit-node-match-p should too.

Does this solve an actual problem you've seen somewhere (and if so,
which problem), or does it solve a potential problem that didn't
actually happen yet?

Also, does CHECK_TS_NODE allow nil or does it currently signal an
error?

bug#74612: 30.0.90; [PATCH] Allow passing nil to treesit-node-match-p

[Yuan Fu]

> On Nov 29, 2024, at 11:22 PM, Eli Zaretskii <eliz@gnu.org> wrote:
> 
>> From: Yuan Fu <casouri@gmail.com>
>> Date: Fri, 29 Nov 2024 16:38:22 -0800
>> 
>> Eli, are you ok with this patch on emacs-30? Almost all tree-sitter node functions allows user to pass nil for NODE, so treesit-node-match-p should too.
> 
> Does this solve an actual problem you've seen somewhere (and if so,
> which problem), or does it solve a potential problem that didn't
> actually happen yet?

I encountered this problem (more like an inconvenience, since without this feature I’d need to test for nullness before passing a node to treesit-node-match-p) when using treesit-node-match-p for some new code I’m writing for master branch. Since (treesit-node-match-p node “type name”) is easier to write and shorter than (equal (treesit-node-type node) “type name”) or (string-match-p “type name regex” (treesit-node-type node)), I’ve been using it liberally in new code :)

> 
> Also, does CHECK_TS_NODE allow nil or does it currently signal an
> error?

CHECK_TS_NODE currently signals an error. All the tree-sitter node functions that accepts nil has the "if (NILP (node)) return Qnil” line before CHECK_TS_NODE (or treesit_check_node)

If you’re not too comfortable with the change, we can apply it to master. It wouldn’t create any backward-incompatibility since this change makes treesit-node-match-p more lenient on its argument, not stricter. The drawback is treesit-node-match-p will be a bit more annoying to use.

Yuan

bug#74612: 30.0.90; [PATCH] Allow passing nil to treesit-node-match-p

[Eli Zaretskii]

> From: Yuan Fu <casouri@gmail.com>
> Date: Sun, 1 Dec 2024 00:31:05 -0800
> Cc: 74612@debbugs.gnu.org
> 
> 
> 
> > On Nov 29, 2024, at 11:22 PM, Eli Zaretskii <eliz@gnu.org> wrote:
> > 
> >> From: Yuan Fu <casouri@gmail.com>
> >> Date: Fri, 29 Nov 2024 16:38:22 -0800
> >> 
> >> Eli, are you ok with this patch on emacs-30? Almost all tree-sitter node functions allows user to pass nil for NODE, so treesit-node-match-p should too.
> > 
> > Does this solve an actual problem you've seen somewhere (and if so,
> > which problem), or does it solve a potential problem that didn't
> > actually happen yet?
> 
> I encountered this problem (more like an inconvenience, since without this feature I’d need to test for nullness before passing a node to treesit-node-match-p) when using treesit-node-match-p for some new code I’m writing for master branch. Since (treesit-node-match-p node “type name”) is easier to write and shorter than (equal (treesit-node-type node) “type name”) or (string-match-p “type name regex” (treesit-node-type node)), I’ve been using it liberally in new code :)
> 
> > 
> > Also, does CHECK_TS_NODE allow nil or does it currently signal an
> > error?
> 
> CHECK_TS_NODE currently signals an error. All the tree-sitter node functions that accepts nil has the "if (NILP (node)) return Qnil” line before CHECK_TS_NODE (or treesit_check_node)
> 
> If you’re not too comfortable with the change, we can apply it to master. It wouldn’t create any backward-incompatibility since this change makes treesit-node-match-p more lenient on its argument, not stricter. The drawback is treesit-node-match-p will be a bit more annoying to use.

OK, please install on emacs-30, and thanks.