From 4ea2a9c459134301dac3456ef38f17acca2326d5 Mon Sep 17 00:00:00 2001
From: mertcandav <mertcandav@outlook.com>
Date: Thu, 17 Oct 2024 19:40:25 +0300
Subject: [PATCH] compiler: improve the --opt-str optimization flag for the
 string concatenations

---
 src/julec/obj/cxx/expr.jule | 25 ++++++++++++++
 src/julec/opt/expr.jule     | 66 +++++++++++++++++++++++++++++++++++--
 src/julec/opt/model.jule    |  4 +++
 3 files changed, 92 insertions(+), 3 deletions(-)

diff --git a/src/julec/obj/cxx/expr.jule b/src/julec/obj/cxx/expr.jule
index f662408d7..41e294e15 100644
--- a/src/julec/obj/cxx/expr.jule
+++ b/src/julec/obj/cxx/expr.jule
@@ -34,6 +34,7 @@ enum compExprModel: type {
 	&opt::UnsafeDerefExprModel,
 	&opt::UnsafeCastingExprModel,
 	&opt::FnCallIgnoreExceptionalExprModel,
+	&opt::StrConcatExprModel,
 }
 
 struct exprCoder {
@@ -1653,6 +1654,28 @@ impl exprCoder {
 		}
 	}
 
+	fn strConcat(mut &self, mut m: &opt::StrConcatExprModel) {
+		self.oc.write("({ " + typeCoder.Int + " size = 0; ")
+		for i, p in m.Parts {
+			ident := "part" + conv::Itoa(i)
+			self.oc.write(typeCoder.Str + " ")
+			self.oc.write(ident)
+			self.oc.write(" = std::move(")
+			self.possibleRefExpr(p)
+			self.oc.write("); size += ")
+			self.oc.write(ident)
+			self.oc.write(".len(); ")
+		}
+		self.oc.write("jule::Slice<" + typeCoder.U8 + "> buf; buf.alloc_new(0, size); ")
+		for i in m.Parts {
+			ident := "part" + conv::Itoa(i)
+			self.oc.write("buf.append(")
+			self.oc.write(ident)
+			self.oc.write("); ")
+		}
+		self.oc.write(typeCoder.Str + " result; result._len = buf._len; result.buffer = std::move(buf.data); result._slice = buf._slice; std::move(result); })")
+	}
+
 	fn model(mut &self, mut m: compExprModel) {
 		match type m {
 		| str:
@@ -1749,6 +1772,8 @@ impl exprCoder {
 			self.unsafeCasting((&opt::UnsafeCastingExprModel)(m))
 		| &opt::FnCallIgnoreExceptionalExprModel:
 			self.funcCall((&opt::FnCallIgnoreExceptionalExprModel)(m).Base, true)
+		| &opt::StrConcatExprModel:
+			self.strConcat((&opt::StrConcatExprModel)(m))
 		|:
 			self.oc.write("<unimplemented_expression_model>")
 		}
diff --git a/src/julec/opt/expr.jule b/src/julec/opt/expr.jule
index 93aa4bc22..dd588ef15 100644
--- a/src/julec/opt/expr.jule
+++ b/src/julec/opt/expr.jule
@@ -129,7 +129,7 @@ impl exprOptimizer {
 		ret true
 	}
 
-	fn strCond(self, mut m: &sema::BinaryExprModel): bool {
+	fn strCond(self, mut &m: &sema::BinaryExprModel): bool {
 		lp := m.Left.Type.Prim()
 		if lp == nil || !lp.IsStr() {
 			ret false
@@ -186,6 +186,65 @@ impl exprOptimizer {
 		ret true
 	}
 
+	// Tries to optimize string concatenations and reports whether applied.
+	fn strConcat(self, mut &m: &sema::BinaryExprModel): bool {
+		lp := m.Left.Type.Prim()
+		if lp == nil || !lp.IsStr() {
+			ret false
+		}
+		if m.Op.Id != token::Id.Plus {
+			ret false
+		}
+
+		// Check the left operand of the binary expression to determine we need to optimize it specifically.
+		// Use the left perand because of CAST structure.
+		// See the variable step and function body documentation about binary expressions.
+		match type m.Left.Model {
+		| &sema::BinaryExprModel:
+			// the left operand expression is binary expression
+			// the expression is like: "x + y + ..."
+			// we can optimize it
+			break
+		|:
+			// expression is like: "x + y"
+			// there is no need to optimize it
+			ret false
+		}
+
+		mut model := new(StrConcatExprModel)
+		let mut step: fn(m: sema::ExprModel)
+		// Point to the step variable in the closure. Otherwise it will be copied,
+		// so the step function remains as nil, which is causes nil dereferencing.
+		&_step := step
+		step = fn(m: sema::ExprModel) {
+			match type m {
+			| &sema::BinaryExprModel:
+				// First, handle the left operand.
+				// Parts should be in same addition order of the source code.
+				// By operator precedence and CAST production,
+				// the first node of the CAST is always the last binary expression.
+				// For exampple:
+				//   _ = "foo" + x + "bar" + "baz"
+				//                         ^
+				//  In the expression above, the first binary expression node will be pointed one.
+				//  So handle `"foo" + x + "bar"` expression first, then the right operand `+ "baz"`.
+				//  So, recursively handle the left binary expression operands,
+				//  and push the expression models by left-to-right order.
+				unsafe { _step((&sema::BinaryExprModel)(m).Left.Model) }
+				unsafe { _step((&sema::BinaryExprModel)(m).Right.Model) }
+			|:
+				model.Parts = append(model.Parts, m)
+			}
+		}
+		// Start handling from the last binary expression m.
+		// See documentation of the step function about binary expressions.
+		step(m)
+
+		mut anyModel := any(model)
+		*self.model = unsafe { *(*sema::ExprModel)(&anyModel) }
+		ret true
+	}
+
 	fn tryNeutralElement1(self, mut &m: &sema::BinaryExprModel, mut c: &constant::Const, mut &nc: &sema::OperandExprModel): bool {
 		if c.IsStr() { // Constant is string, check for string optimizations.
 			if !Str {
@@ -361,8 +420,9 @@ impl exprOptimizer {
 		}
 
 		if Str {
-			if self.strCond(m) {
-				ret
+			match {
+			| self.strCond(m)
+			| self.strConcat(m):
 			}
 		}
 
diff --git a/src/julec/opt/model.jule b/src/julec/opt/model.jule
index 980879830..f18288501 100644
--- a/src/julec/opt/model.jule
+++ b/src/julec/opt/model.jule
@@ -14,6 +14,10 @@ struct FnCallIgnoreExceptionalExprModel {
 	Base: &sema::FnCallExprModel
 }
 
+struct StrConcatExprModel {
+	Parts: []sema::ExprModel
+}
+
 struct SwapExprModel {
 	Left:  &sema::Data
 	Right: &sema::Data