bug#73662: [PATCH] Anchor to first sibling unless bol-sibling found

bug#73662: [PATCH] Anchor to first sibling unless bol-sibling found

[Jørgen Kvalsvik]

[-- Attachment #1: Type: text/plain, Size: 990 bytes --]

Tags: patch


Indenting would give up when initializer lists and similar constructs
were broken up over multiple lines where the first item was not also
beginning-of-line.  Items should anchor on the first sibling (like with
function calls and init-lists in general). The search for a sibling to
anchor on should stop at the first sibling at bol or the first sibling
in the (sub)tree.

This brings indent up to c-mode parity for code like:

return { x1, x2, ...,
         y1, x2, ... };


In GNU Emacs 31.0.50 (build 1, x86_64-pc-linux-gnu) of 2024-10-06 built
 on ada
Repository revision: c8e5f2ee9f6155ef545c2fe8ddf702c4f16a6eea
Repository branch: master
System Description: Debian GNU/Linux 12 (bookworm)

Configured using:
 'configure --with-x-toolkit=no --with-xpm=ifavailable
 --with-jpeg=ifavailable --with-png=ifavailable --with-gif=ifavailable
 --with-tiff=ifavailable --with-gnutls=ifavailable --with-tree-sitter
 --with-native-compilation --prefix=/home/j/src/emacs/build/root'


[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #2: 0001-Anchor-to-first-sibling-unless-bol-sibling-found.patch --]
[-- Type: text/patch, Size: 3682 bytes --]

From 0bf96b008b57298ffcc63b9bb8222358ce966c21 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?J=C3=B8rgen=20Kvalsvik?= <j@lambda.is>
Date: Fri, 4 Oct 2024 21:38:33 +0200
Subject: [PATCH] Anchor to first sibling unless bol-sibling found

The behavior differed between c-mode/c++-mode and *-ts-mode for
initializer lists where the first element was not at beginning-of-line.
The anchor-prev-sibling function gave up and returned nil, but it should
(probably) anchor on the first element in the initializer list, such as
this:

return { v1, v2, ...,
         y1, y2, ... };

c-ts-mode behaved better and figured out how to align, but I added a
test for a similar compound literal to prevent regressions.

	* lisp/progmodes/c-ts-mode.el (c-ts-mode--anchor-prev-sibling):
	Anchor at first sibling unless bol is found.

	* test/lisp/progmodes/c-ts-mode-resources/indent.erts: New
	initializer list and compound literal test.
---
 lisp/progmodes/c-ts-mode.el                   |  9 +++--
 .../progmodes/c-ts-mode-resources/indent.erts | 39 +++++++++++++++++++
 2 files changed, 45 insertions(+), 3 deletions(-)

diff --git a/lisp/progmodes/c-ts-mode.el b/lisp/progmodes/c-ts-mode.el
index 303c994637c..3f6a7422aa0 100644
--- a/lisp/progmodes/c-ts-mode.el
+++ b/lisp/progmodes/c-ts-mode.el
@@ -338,10 +338,13 @@ c-ts-mode--anchor-prev-sibling
           ;; If the start of the previous sibling isn't at the
           ;; beginning of a line, something's probably not quite
           ;; right, go a step further. (E.g., comment after a
-          ;; statement.)
+          ;; statement.)  If the previous sibling is the first named
+          ;; node then anchor to that, e.g. when returning an aggregate
+          ;; and starting the items on the same line as {.
           (_ (goto-char (treesit-node-start prev-sibling))
-             (if (looking-back (rx bol (* whitespace))
-                               (line-beginning-position))
+             (if (or (looking-back (rx bol (* whitespace))
+                                   (line-beginning-position)))
+                     (null (treesit-node-prev-sibling prev-sibling t))
                  (setq continue nil)
                (setq prev-sibling
                      (treesit-node-prev-sibling prev-sibling)))))))
diff --git a/test/lisp/progmodes/c-ts-mode-resources/indent.erts b/test/lisp/progmodes/c-ts-mode-resources/indent.erts
index 599173832b5..a13a74cf8b3 100644
--- a/test/lisp/progmodes/c-ts-mode-resources/indent.erts
+++ b/test/lisp/progmodes/c-ts-mode-resources/indent.erts
@@ -208,6 +208,21 @@ int main()
 }
 =-=-=
 
+Name: Return Compund Literal
+
+=-=
+struct pair { int fst, snd; };
+struct pair
+make_pair(int long_identifier_a[], int long_identifier_b[],
+	  int offset_a, int offset_b)
+{
+  int base_offset = 10;
+  return (struct pair) { long_identifier_a[base_offset + offset_b],
+                         long_identifier_b[base_offset + offset_b] };
+}
+
+=-=-=
+
 Name: Switch-Case statement
 
 =-=
@@ -486,6 +501,30 @@ namespace A {
 }
 =-=-=
 
+Name: Return Aggregate Initialized Struct
+
+=-=
+struct pair { int x, y; }
+pair
+make_pair(int long_identifier_a[], int long_identifier_b[],
+          int offset_a, int offset_b)
+{
+  int base_offset = 10;
+  return { long_identifier_a[base_offset + offset_b],
+       long_identifier_b[base_offset + offset_b] };
+}
+=-=
+struct pair { int x, y; }
+pair
+make_pair(int long_identifier_a[], int long_identifier_b[],
+          int offset_a, int offset_b)
+{
+  int base_offset = 10;
+  return { long_identifier_a[base_offset + offset_b],
+           long_identifier_b[base_offset + offset_b] };
+}
+=-=-=
+
 Code:
   (lambda ()
     (c-ts-mode)
-- 
2.39.5

bug#73662: [PATCH] Anchor to first sibling unless bol-sibling found

[Eli Zaretskii]

merge 73662 73661
thanks

> Cc: theo@thornhill.no,casouri@gmail.com
> From: Jørgen Kvalsvik <j@lambda.is>
> Date: Sun, 06 Oct 2024 16:03:44 +0200
> 
> Indenting would give up when initializer lists and similar constructs
> were broken up over multiple lines where the first item was not also
> beginning-of-line.  Items should anchor on the first sibling (like with
> function calls and init-lists in general). The search for a sibling to
> anchor on should stop at the first sibling at bol or the first sibling
> in the (sub)tree.
> 
> This brings indent up to c-mode parity for code like:
> 
> return { x1, x2, ...,
>          y1, x2, ... };

This is an exact duplicate of bug 73661, which you submitted just a
few minutes ago.  So I'm merging them.