From 359e28da67a7838094265af74be8570743b046d6 Mon Sep 17 00:00:00 2001
From: Alexander Medvednikov <alexander@medvednikov.com>
Date: Wed, 25 Mar 2026 16:42:24 +0300
Subject: [PATCH] cgen: optimize string concatenation chains (fixes #276)

---
 vlib/builtin/string.v                         | 27 +++++++++
 vlib/v/gen/c/infix.v                          | 55 +++++++++++++++++++
 ...ring_concat_chain_optimization.c.must_have |  2 +
 .../string_concat_chain_optimization.vv       | 12 ++++
 4 files changed, 96 insertions(+)
 create mode 100644 vlib/v/gen/c/testdata/string_concat_chain_optimization.c.must_have
 create mode 100644 vlib/v/gen/c/testdata/string_concat_chain_optimization.vv

diff --git a/vlib/builtin/string.v b/vlib/builtin/string.v
index 0a337a7d9..435b96e6b 100644
--- a/vlib/builtin/string.v
+++ b/vlib/builtin/string.v
@@ -918,6 +918,33 @@ fn (s string) + (a string) string {
 	return res
 }
 
+// string_plus_many concatenates several strings with a single allocation.
+@[direct_array_access; markused]
+fn string_plus_many(data_len int, input_base &string) string {
+	mut new_len := 0
+	for i := 0; i < data_len; i++ {
+		part := unsafe { input_base[i] }
+		new_len += if part.len > 0 { part.len } else { 0 }
+	}
+	mut res := string{
+		str: unsafe { malloc_noscan(new_len + 1) }
+		len: new_len
+	}
+	mut offset := 0
+	unsafe {
+		for i := 0; i < data_len; i++ {
+			part := input_base[i]
+			part_len := if part.len > 0 { part.len } else { 0 }
+			if part_len > 0 {
+				vmemcpy(res.str + offset, part.str, part_len)
+				offset += part_len
+			}
+		}
+		res.str[new_len] = 0 // V strings are not null terminated, but just in case
+	}
+	return res
+}
+
 // for `s + s2 + s3`, an optimization (faster than string_plus(string_plus(s1, s2), s3))
 @[direct_array_access]
 fn (s string) plus_two(a string, b string) string {
diff --git a/vlib/v/gen/c/infix.v b/vlib/v/gen/c/infix.v
index 3c1160962..dfd66a0ef 100644
--- a/vlib/v/gen/c/infix.v
+++ b/vlib/v/gen/c/infix.v
@@ -1163,7 +1163,62 @@ fn (mut g Gen) gen_interface_is_op(node ast.InfixExpr) {
 
 // infix_expr_arithmetic_op generates code for `+`, `-`, `*`, `/`, and `%`
 // It handles operator overloading when necessary
+fn (mut g Gen) is_string_type(typ ast.Type) bool {
+	return g.unwrap(typ).unaliased_sym.kind == .string
+}
+
+fn (mut g Gen) is_string_concat_infix(node ast.InfixExpr) bool {
+	if node.op != .plus {
+		return false
+	}
+	left_type := g.type_resolver.get_type_or_default(node.left, node.left_type)
+	right_type := g.type_resolver.get_type_or_default(node.right, node.right_type)
+	return g.is_string_type(left_type) && g.is_string_type(right_type)
+}
+
+fn (mut g Gen) collect_string_concat_parts(expr ast.Expr, mut parts []ast.Expr) {
+	match expr {
+		ast.InfixExpr {
+			if g.is_string_concat_infix(expr) {
+				g.collect_string_concat_parts(expr.left, mut parts)
+				g.collect_string_concat_parts(expr.right, mut parts)
+				return
+			}
+		}
+		ast.ParExpr {
+			g.collect_string_concat_parts(expr.expr, mut parts)
+			return
+		}
+		else {}
+	}
+	parts << expr
+}
+
+fn (mut g Gen) gen_string_concat_many(node ast.InfixExpr) bool {
+	if !g.is_string_concat_infix(node) {
+		return false
+	}
+	mut parts := []ast.Expr{}
+	g.collect_string_concat_parts(ast.Expr(node), mut parts)
+	if parts.len < 3 {
+		return false
+	}
+	g.write('builtin__string_plus_many(${parts.len}, _MOV((string[${parts.len}]){')
+	for i, part in parts {
+		part_type := g.type_resolver.get_type_or_default(part, part.type())
+		g.expr_with_cast(part, part_type, ast.string_type)
+		if i < parts.len - 1 {
+			g.write(', ')
+		}
+	}
+	g.write('}))')
+	return true
+}
+
 fn (mut g Gen) infix_expr_arithmetic_op(node ast.InfixExpr) {
+	if g.gen_string_concat_many(node) {
+		return
+	}
 	left_type := g.resolved_expr_type(node.left, node.left_type)
 	right_type := g.resolved_expr_type(node.right, node.right_type)
 	$if trace_ci_fixes ? {
diff --git a/vlib/v/gen/c/testdata/string_concat_chain_optimization.c.must_have b/vlib/v/gen/c/testdata/string_concat_chain_optimization.c.must_have
new file mode 100644
index 000000000..2897bea31
--- /dev/null
+++ b/vlib/v/gen/c/testdata/string_concat_chain_optimization.c.must_have
@@ -0,0 +1,2 @@
+builtin__string_plus_many(3, _MOV((string[3]){a, b, c}))
+builtin__string_plus_many(4, _MOV((string[4]){a, b, c, d}))
diff --git a/vlib/v/gen/c/testdata/string_concat_chain_optimization.vv b/vlib/v/gen/c/testdata/string_concat_chain_optimization.vv
new file mode 100644
index 000000000..cbc0aca3e
--- /dev/null
+++ b/vlib/v/gen/c/testdata/string_concat_chain_optimization.vv
@@ -0,0 +1,12 @@
+fn join_three(a string, b string, c string) string {
+	return a + b + c
+}
+
+fn join_four(a string, b string, c string, d string) string {
+	return a + b + c + d
+}
+
+fn main() {
+	assert join_three('a', 'b', 'c') == 'abc'
+	assert join_four('a', 'b', 'c', 'd') == 'abcd'
+}
-- 
2.39.5