native: implement for in string for amd64 (#24613)

Author: Eliyaan

vlib/v/gen/native/amd64.v

@@ -212,6 +212,10 @@ fn (mut c Amd64) cmp(reg Amd64Register, size Size, val i64) {
 	c.g.println('cmp ${reg}, ${val}')
 }
 
+fn (mut c Amd64) cmp_reg2(reg Register, reg2 Register) {
+	c.cmp_reg(reg as Amd64Register, reg2 as Amd64Register)
+}
+
 // `cmp rax, rbx`
 fn (mut c Amd64) cmp_reg(reg Amd64Register, reg2 Amd64Register) {
 	match reg {
@@ -1101,8 +1105,12 @@ fn (mut c Amd64) push(r Register) {
 		c.g.write8(0x50 + i32(reg) - 8)
 	}
 	c.is_16bit_aligned = !c.is_16bit_aligned
-	c.g.println('push ${reg}')
 	c.g.stack_depth++
+	c.g.println('push ${reg}; stack_depth:${c.g.stack_depth}')
+}
+
+fn (mut c Amd64) pop2(r Register) {
+	c.pop(r as Amd64Register)
 }
 
 pub fn (mut c Amd64) pop(reg Amd64Register) {
@@ -1111,8 +1119,8 @@ pub fn (mut c Amd64) pop(reg Amd64Register) {
 	}
 	c.g.write8(0x58 + i32(reg) % 8)
 	c.is_16bit_aligned = !c.is_16bit_aligned
-	c.g.println('pop ${reg}')
 	c.g.stack_depth--
+	c.g.println('pop ${reg} ; stack_depth:${c.g.stack_depth}')
 }
 
 pub fn (mut c Amd64) sub8(reg Amd64Register, val i32) {
@@ -3620,7 +3628,7 @@ pub fn (mut c Amd64) allocate_var(name string, size i32, initial_val Number) i32
 	}
 
 	// println('allocate_var(size=$size, initial_val=$initial_val)')
-	c.g.println('mov [rbp-${int(n).hex2()}], ${initial_val} ; Allocate var `${name}`')
+	c.g.println('mov [rbp-${int(n).hex2()}], ${initial_val} ; Allocate var `${name}` size: ${size}')
 	return c.g.stack_var_pos
 }
 

vlib/v/gen/native/arm64.v

@@ -535,6 +535,14 @@ pub fn (mut c Arm64) add_reg2(r Register, r2 Register) {
 	panic('Arm64.add_reg2() not implemented')
 }
 
+fn (mut c Arm64) pop2(r Register) {
+	panic('Arm64.pop2() not implemented')
+}
+
+fn (mut c Arm64) cmp_reg2(reg Register, reg2 Register) {
+	panic('Arm64.add_reg2() not implemented')
+}
+
 fn (mut c Arm64) create_string_struct(typ ast.Type, name string, str string) {
 	panic('Arm64.add_reg2() not implemented')
 }

vlib/v/gen/native/blacklist.v

@@ -46,7 +46,12 @@ const whitelist = {
 	'string.is_capital':    false
 	'string.is_ascii':      false
 	'string.is_identifier': false
-	//	'string.is_blank':      false need for in
+	'string.is_blank':      false
+	'string.indent_width':  false
+	'string.index_u8':      false
+	'string.last_index':    true
+	'string.last_index_u8': false
+	'string.contains_u8':   false
 }
 
 fn (g &Gen) is_blacklisted(name string, is_builtin bool) bool {

vlib/v/gen/native/gen.v

@@ -95,6 +95,7 @@ mut:
 	cmp_to_stack_top(r Register)
 	cmp_var_reg(var Var, reg Register, config VarConfig)
 	cmp_var(var Var, val i32, config VarConfig)
+	cmp_reg2(reg Register, reg2 Register)
 	cmp_zero(reg Register)
 	convert_bool_to_string(r Register)
 	convert_int_to_string(a Register, b Register)
@@ -131,6 +132,7 @@ mut:
 	mov64(r Register, val Number)
 	movabs(reg Register, val i64)
 	patch_relative_jmp(pos i32, addr i64)
+	pop2(r Register)
 	prefix_expr(node ast.PrefixExpr)
 	push(r Register)
 	ret()

vlib/v/gen/native/stmt.c.v

@@ -255,14 +255,14 @@ fn (mut g Gen) for_stmt(node ast.ForStmt) {
 }
 
 fn (mut g Gen) for_in_stmt(node ast.ForInStmt) { // Work on that
+	main_reg := g.code_gen.main_reg()
 	if node.is_range {
 		g.println('; for ${node.val_var} in range {')
 		// for a in node.cond .. node.high {
 
 		i := g.code_gen.allocate_var(node.val_var, 8, i64(0)) // iterator variable
 		g.println('; evaluate node.cond for lower bound:')
 		g.expr(node.cond) // outputs the lower loop bound (initial value) to the main reg
-		main_reg := g.code_gen.main_reg()
 		g.println('; move the result to i')
 		g.code_gen.mov_reg_to_var(LocalVar{i, ast.i64_type_idx, node.val_var}, main_reg) // i = node.cond // initial value
 
@@ -299,11 +299,79 @@ fn (mut g Gen) for_in_stmt(node ast.ForInStmt) { // Work on that
 		g.labels.addrs[end_label] = g.pos()
 		g.println('; label ${end_label} (end_label)')
 		g.println('; for ${node.val_var} in range }')
+	} else if node.kind == .string {
+		g.println('; for ${node.val_var} in string {')
+		// for c in my_string {
+
+		key_var := if node.key_var == '' { 'i' } else { node.key_var }
+		i := g.code_gen.allocate_var(key_var, 8, i64(0)) // iterator variable
+		c := g.code_gen.allocate_var(node.val_var, 1, i64(0)) // char variable
+
+		g.expr(node.cond) // get the address of the string variable
+		g.code_gen.mov_deref(Amd64Register.rdx, main_reg, ast.charptr_type)
+		g.println('; push address of the string chars')
+		g.code_gen.push(Amd64Register.rdx) // address of the string
+		g.code_gen.add(main_reg, g.get_field_offset(ast.string_type, 'len'))
+		g.println('; push address of the len:')
+		g.code_gen.push(main_reg) // address of the len
+
+		start := g.pos() // label-begin:
+
+		g.println('; check iterator against upper loop bound')
+		g.code_gen.mov_var_to_reg(main_reg, LocalVar{i, ast.i64_type_idx, key_var})
+		g.println('; pop address of the len:')
+		g.code_gen.pop2(Amd64Register.rdx)
+		g.println('; push address of the len:')
+		g.code_gen.push(Amd64Register.rdx) // len
+		g.code_gen.mov_deref(Amd64Register.rdx, Amd64Register.rdx, ast.int_type)
+		g.code_gen.cmp_reg2(main_reg, Amd64Register.rdx)
+		jump_addr := g.code_gen.cjmp(.jge) // leave loop i >= len
+
+		g.println('; pop address of the len:')
+		g.code_gen.pop2(Amd64Register.rdx) // len
+		g.println('; pop address of the string chars')
+		g.code_gen.pop2(Amd64Register.rax) // address of the string
+		g.println('; push address of the string chars')
+		g.code_gen.push(Amd64Register.rax)
+		g.println('; push address of the len:')
+		g.code_gen.push(Amd64Register.rdx) // len
+
+		g.code_gen.mov_var_to_reg(Amd64Register.rdx, LocalVar{i, ast.i64_type_idx, key_var})
+		g.code_gen.add_reg2(Amd64Register.rax, Amd64Register.rdx)
+		g.code_gen.mov_deref(Amd64Register.rax, Amd64Register.rax, ast.u8_type_idx)
+		g.code_gen.mov_reg_to_var(LocalVar{c, ast.u8_type_idx, node.val_var}, Amd64Register.rax) // store the char
+
+		end_label := g.labels.new_label()
+		g.labels.patches << LabelPatch{
+			id:  end_label
+			pos: jump_addr
+		}
+		g.println('; jump to label ${end_label} (end_label)')
+
+		start_label := g.labels.new_label() // used for continue
+		g.labels.branches << BranchLabel{
+			name:  node.label
+			start: start_label
+			end:   end_label
+		}
+		g.stmts(node.stmts) // writes the actual body of the loop
+
+		g.labels.addrs[start_label] = g.pos() // used for continue (continue: jump before the inc)
+		g.println('; label ${start_label} (continue_label)')
+
+		g.code_gen.inc_var(LocalVar{i, ast.i64_type_idx, key_var})
+		g.labels.branches.pop()
+		g.code_gen.jmp_back(start) // loops
+
+		g.labels.addrs[end_label] = g.pos()
+		g.code_gen.pop2(Amd64Register.rdx) // len
+		g.code_gen.pop2(Amd64Register.rax) // address of the string
+		g.println('; label ${end_label} (end_label)')
+		g.println('; for ${node.val_var} in string }')
 		/*
 	} else if node.kind == .array {
 	} else if node.kind == .array_fixed {
 	} else if node.kind == .map {
-	} else if node.kind == .string {
 	} else if node.kind == .struct {
 	} else if it.kind in [.array, .string] || it.cond_type.has_flag(.variadic) {
 	} else if it.kind == .map {

vlib/v/gen/native/tests/builtin.vv

@@ -22,7 +22,6 @@ assert !a.is_ascii()
 	assert '_9'.is_identifier() == true
 	assert 'a 9'.is_identifier() == false
 	assert 't'.is_identifier() == true
-/*
 	assert ''.is_blank()
 	assert ' '.is_blank()
 	assert ' \t'.is_blank()
@@ -48,18 +47,18 @@ assert !a.is_ascii()
 	assert 'abc'.index_u8(`A`) == -1
 	assert 'abc'.index_u8(`B`) == -1
 	assert 'abc'.index_u8(`C`) == -1
+/*
 	assert 'abcabca'.last_index('ca')? == 5
 	assert 'abcabca'.last_index('ab')? == 3
 	assert 'abcabca'.last_index('b')? == 4
 	assert 'Zabcabca'.last_index('Z')? == 0
 	x := 'Zabcabca'.last_index('Y')
 	assert x == none
-	// TODO: `assert 'Zabcabca'.index_last('Y') == none` is a cgen error, 2023/12/04
+*/
 	assert 'abcabca'.last_index_u8(`a`) == 6
 	assert 'abcabca'.last_index_u8(`c`) == 5
 	assert 'abcabca'.last_index_u8(`b`) == 4
 	assert 'Zabcabca'.last_index_u8(`Z`) == 0
-	//
 	assert 'abc'.last_index_u8(`d`) == -1
 	assert 'abc'.last_index_u8(`A`) == -1
 	assert 'abc'.last_index_u8(`B`) == -1
@@ -69,6 +68,7 @@ assert !a.is_ascii()
 	assert 'abc abca'.contains_u8(`c`)
 	assert 'abc abca'.contains_u8(` `)
 	assert !'abc abca'.contains_u8(`A`)
+/*
 	assert 'Abcd'.camel_to_snake() == 'abcd'
 	assert 'aBcd'.camel_to_snake() == 'a_bcd'
 	assert 'AAbb'.camel_to_snake() == 'aa_bb'

vlib/v/gen/native/tests/linux.vv

@@ -0,0 +1,21 @@
+
+fn test_for_c_in_string() {
+	$if windows {
+		println('0')
+		println('97')
+		println('1')
+		println('98')
+		println('2')
+		println('99')
+	} $else {
+		s := 'abc'
+		for i, c in s {
+			println(i)
+			println(int(c))
+		}
+	}
+}
+
+fn main() {
+	test_for_c_in_string()
+}

vlib/v/gen/native/tests/linux.vv.out

@@ -0,0 +1,6 @@
+0
+97
+1
+98
+2
+99

vlib/v/gen/native/tests/native_test.v

@@ -49,7 +49,7 @@ fn test_native() {
 				continue
 			}
 		}
-		if test == 'fibonacci_native.vv' {
+		if test == 'fibonacci_native.vv' || test.contains('linux') {
 			if user_os == 'windows' {
 				println('>>> SKIPPING ${test} on windows for now')
 				continue