gasm-sdk/format/format_test.go

// Copyright (c) 2026 Petr Balvín <opensource@petrbalvin.org> (https://petrbalvin.org)
// SPDX-License-Identifier: BSD-3-Clause

package format

import (
	"os"
	"slices"
	"strings"
	"testing"

	"sourcedock.dev/petrbalvin/gasm-devkit/ast"
	"sourcedock.dev/petrbalvin/gasm-devkit/lexer"
	"sourcedock.dev/petrbalvin/gasm-devkit/parser"
	"sourcedock.dev/petrbalvin/gasm-devkit/token"
)

func TestGolden(t *testing.T) {
	in := "#include \"textflag.h\"\n" +
		"\n" +
		"TEXT ·f(SB), NOSPLIT, $0\n" +
		"MOVQ swin_base+0(FP), SI\n" +
		"LEAQ (SI)(BX*4), R9\n" +
		"ANDQ $-8, R10\n" +
		"VFMADD231PD Z14, Z12, Z10\n" +
		"RET\n"

	want := "#include \"textflag.h\"\n" +
		"\n" +
		"TEXT ·f(SB), NOSPLIT, $0\n" +
		"\tMOVQ        swin_base+0(FP), SI\n" +
		"\tLEAQ        (SI)(BX*4), R9\n" +
		"\tANDQ        $-8, R10\n" +
		"\tVFMADD231PD Z14, Z12, Z10\n" +
		"\tRET\n"

	got := Source(in)
	if got != want {
		t.Fatalf("formatting mismatch:\n--- got ---\n%q\n--- want ---\n%q", got, want)
	}
}

// TestDocCommentIndent checks that a doc comment preceding a TEXT directive
// sits at column 0 even when another function (ending in RET) precedes it;
// the RET must terminate the previous body for indentation purposes.
func TestDocCommentIndent(t *testing.T) {
	in := "#include \"textflag.h\"\n" +
		"\n" +
		"// func first()\n" +
		"TEXT ·first(SB), NOSPLIT, $0\n" +
		"XORQ AX, AX\n" +
		"RET\n" +
		"\n" +
		"// func second()\n" +
		"TEXT ·second(SB), NOSPLIT, $0\n" +
		"RET\n"

	want := "#include \"textflag.h\"\n" +
		"\n" +
		"// func first()\n" +
		"TEXT ·first(SB), NOSPLIT, $0\n" +
		"\tXORQ AX, AX\n" +
		"\tRET\n" +
		"\n" +
		"// func second()\n" +
		"TEXT ·second(SB), NOSPLIT, $0\n" +
		"\tRET\n"

	got := Source(in)
	if got != want {
		t.Fatalf("formatting mismatch:\n--- got ---\n%q\n--- want ---\n%q", got, want)
	}
	// Body comments stay indented.
	body := "#include \"textflag.h\"\nTEXT ·f(SB), NOSPLIT, $0\n// inside the body\nXORQ AX, AX\nRET\n"
	gotBody := Source(body)
	if !strings.Contains(gotBody, "\t// inside the body\n") {
		t.Fatalf("body comment must stay indented:\n%q", gotBody)
	}
}

// TestBlankLines checks the blank-line canonicalisation: exactly one blank
// line before a new block (a label, or TEXT/GLOBL), runs of blanks collapsed
// to one, and no blank forced after TEXT, between stacked labels, or at the
// top of the file.  Leading comments belong to the block they precede.
func TestBlankLines(t *testing.T) {
	in := "#include \"textflag.h\"\n" +
		"TEXT ·f(SB), NOSPLIT, $0\n" +
		"first:\n" + // first label: no blank after TEXT
		"XORQ AX, AX\n" +
		"JMP next\n" + // unlabelled glue: fmt inserts a blank before next:
		"next:\n" +
		"stacked:\n" + // stacked labels share an address: no blank between
		"INCQ AX\n" +
		"\n" +
		"\n" + // two blanks collapse to one
		"// separated block\n" + // comment belongs to the label below
		"later:\n" +
		"RET\n" +
		"// func g()\n" + // doc comment: blank goes before it
		"TEXT ·g(SB), NOSPLIT, $0\n" +
		"RET\n" +
		"GLOBL ·mask(SB), RODATA, $8\n" + // blank before GLOBL…
		"DATA ·mask+0(SB)/4, $1\n" + // …but not before DATA
		"\n" +
		"\n" +
		"\n" // trailing blanks dropped

	want := "#include \"textflag.h\"\n" +
		"\n" +
		"TEXT ·f(SB), NOSPLIT, $0\n" +
		"first:\n" +
		"\tXORQ AX, AX\n" +
		"\tJMP  next\n" +
		"\n" +
		"next:\n" +
		"stacked:\n" +
		"\tINCQ AX\n" +
		"\n" +
		"\t// separated block\n" + // body comment before a label stays indented
		"later:\n" +
		"\tRET\n" +
		"\n" +
		"// func g()\n" +
		"TEXT ·g(SB), NOSPLIT, $0\n" +
		"\tRET\n" +
		"\n" +
		"GLOBL ·mask(SB), RODATA, $8\n" +
		"DATA ·mask+0(SB)/4, $1\n"

	got := Source(in)
	if got != want {
		t.Fatalf("formatting mismatch:\n--- got ---\n%q\n--- want ---\n%q", got, want)
	}
	if again := Source(got); again != got {
		t.Fatalf("not idempotent:\n%q", again)
	}
}

func TestOperandSpacing(t *testing.T) {
	cases := map[string]string{
		"4(SI)":           "4(SI)",
		"(SI)(BX*4)":      "(SI)(BX*4)",
		"$-8":             "$-8",
		"$0x80020100":     "$0x80020100",
		"swin_base+0(FP)": "swin_base+0(FP)",
		"mask24<>(SB)":    "mask24<>(SB)",
		"·idx16+0(SB)/4":  "·idx16+0(SB)/4",
		"NOSPLIT|DUPOK":   "NOSPLIT|DUPOK",
	}
	for in, want := range cases {
		toks := lexOperands(in)
		if got := renderOps(toks); got != want {
			t.Errorf("renderOps(%q) = %q, want %q", in, got, want)
		}
	}
}

// TestVectorBracketSpacing pins the square-bracket operand forms of the
// arm64 and loong64 vector syntaxes.  The lexer emits '[' and ']' as Illegal
// tokens carrying their spelling, and renderOps must glue them back exactly
// where they were: a register list and an element selector are load-bearing
// operands the assembler reads out of the operand text, so no bracket may be
// dropped, and the canonical spelling inside the brackets is tight.
func TestVectorBracketSpacing(t *testing.T) {
	cases := map[string]string{
		// Register lists of one to four registers.
		"[V21.B16]":                            "[V21.B16]",
		"[V17.B16, V18.B16]":                   "[V17.B16, V18.B16]",
		"[V18.D1, V19.D1, V20.D1]":             "[V18.D1, V19.D1, V20.D1]",
		"[V14.B16, V15.B16, V16.B16, V17.B16]": "[V14.B16, V15.B16, V16.B16, V17.B16]",
		// Element selectors.
		"V31.B[15]":            "V31.B[15]",
		"V19.S[0]":             "V19.S[0]",
		"V1.D[1]":              "V1.D[1]",
		"V11.B[11], V16.B[12]": "V11.B[11], V16.B[12]",
		// Lists beside address operands, on either side.
		"32(R1), [V2.B16, V3.B16]": "32(R1), [V2.B16, V3.B16]",
		"[V2.S4, V3.S4], (R14)":    "[V2.S4, V3.S4], (R14)",
		"(R24), [V18.D1, V19.D1]":  "(R24), [V18.D1, V19.D1]",
		// A spaced spelling canonicalises to the tight one.
		"[ V21.B16 ]": "[V21.B16]",
		"V31.B [15]":  "V31.B[15]",
	}
	for in, want := range cases {
		toks := lexOperands(in)
		if got := renderOps(toks); got != want {
			t.Errorf("renderOps(%q) = %q, want %q", in, got, want)
		}
	}
}

// TestSIMDBracketRoundTrip formats whole functions carrying the bracket
// shapes of the arm64 vector kernels and pins the output byte for byte.  The
// brackets are load-bearing: formatting must not change what the file
// assembles to, so the formatted text keeps every bracket, re-formats to
// itself and still parses cleanly.
func TestSIMDBracketRoundTrip(t *testing.T) {
	in := "#include \"textflag.h\"\n" +
		"\n" +
		"TEXT ·f(SB), NOSPLIT, $0\n" +
		"VDUP V31.B[15], R3\n" +
		"VTBL V22.B16, [V28.B16], V11.B16\n" +
		"VLD1 (R2), [V21.B16]\n" +
		"VMOVQ $0x70, $0x80, V10\n" +
		"RET\n"

	want := "#include \"textflag.h\"\n" +
		"\n" +
		"TEXT ·f(SB), NOSPLIT, $0\n" +
		"\tVDUP  V31.B[15], R3\n" +
		"\tVTBL  V22.B16, [V28.B16], V11.B16\n" +
		"\tVLD1  (R2), [V21.B16]\n" +
		"\tVMOVQ $0x70, $0x80, V10\n" +
		"\tRET\n"

	got := Source(in)
	if got != want {
		t.Fatalf("formatting mismatch:\n--- got ---\n%q\n--- want ---\n%q", got, want)
	}
	if again := Source(got); again != got {
		t.Fatalf("not idempotent:\n%q", again)
	}
	if n := strings.Count(got, "["); n != 3 {
		t.Errorf("output carries %d '[', want 3:\n%s", n, got)
	}
	if _, errs := parser.Parse("in.s", got); len(errs) > 0 {
		t.Errorf("formatted output no longer parses: %v", errs)
	}
}

// TestBracketFormsRoundTrip runs every bracket shape of the vector kernels
// through a full format pass as its own single-instruction function, where
// the canonical form is the line itself indented: formatting must be a no-op
// on each, so no bracket moves, vanishes or gains a space.
func TestBracketFormsRoundTrip(t *testing.T) {
	for _, instr := range []string{
		"VDUP V31.B[15], V18",
		"VDUP V19.S[3], V18.S4",
		"VDUP V1.D[1], V2.D2",
		"VMOV V13.S[0], R20",
		"VMOV V11.B[11], V16.B[12]",
		"VMOV R20, V21.B[2]",
		"VTBL V22.B16, [V28.B16], V11.B16",
		"VTBL V18.B8, [V17.B16, V18.B16], V22.B8",
		"VTBL V31.B8, [V14.B16, V15.B16, V16.B16, V17.B16], V15.B8",
		"VLD1 (R2), [V21.B16]",
		"VLD1 (R24), [V18.D1, V19.D1, V20.D1]",
		"VLD1 (R29), [V14.D1, V15.D1, V16.D1, V17.D1]",
		"VLD1.P 32(R1), [V2.B16, V3.B16]",
		"VLD1R (R1), [V9.B8]",
		"VLD4R (R0), [V0.B8, V1.B8, V2.B8, V3.B8]",
		"VST1 [V2.S4, V3.S4, V4.S4, V5.S4], (R14)",
		"VST1.P [V2.B16], (R1)",
		"VST1.P [V2.B16, V3.B16], 32(R1)",
		"VMOVQ $0x70, $0x80, V10",
	} {
		src := "TEXT ·f(SB), NOSPLIT, $0\n" + instr + "\nRET\n"
		want := "TEXT ·f(SB), NOSPLIT, $0\n\t" + instr + "\n\tRET\n"
		got := Source(src)
		if got != want {
			t.Errorf("formatting %q:\n got  %q\n want %q", instr, got, want)
			continue
		}
		if again := Source(got); again != got {
			t.Errorf("not idempotent for %q:\n%q", instr, again)
		}
		if _, errs := parser.Parse("in.s", got); len(errs) > 0 {
			t.Errorf("formatted output of %q no longer parses: %v", instr, errs)
		}
	}
}

// TestFlagListRoundTrip pins the '|' flag separator and the <ABIInternal>
// marker through a full format pass: the bars the Go toolchain requires and
// the ABI bracket must survive byte for byte, on TEXT and GLOBL alike.
func TestFlagListRoundTrip(t *testing.T) {
	for _, in := range []string{
		"TEXT ·f(SB), NOSPLIT|NOFRAME|DUPOK, $0\n\tRET\n",
		"TEXT ·foo<ABIInternal>(SB), NOSPLIT, $-0-24\n\tRET\n",
		"GLOBL ·mask(SB), RODATA|NOPTR, $8\n",
	} {
		if got := Source(in); got != in {
			t.Fatalf("flag list did not round-trip:\n--- got ---\n%q\n--- want ---\n%q", got, in)
		}
	}
}

// TestCRLFInputIsNormalisedToLF checks that a CRLF file comes out with
// uniform LF endings: a // comment must not carry its line's trailing \r
// into the output.
func TestCRLFInputIsNormalisedToLF(t *testing.T) {
	in := "// func f()\r\nTEXT ·f(SB), NOSPLIT, $0\r\nRET\r\n"
	want := "// func f()\nTEXT ·f(SB), NOSPLIT, $0\n\tRET\n"
	got := Source(in)
	if got != want {
		t.Fatalf("CRLF formatting mismatch:\n--- got ---\n%q\n--- want ---\n%q", got, want)
	}
	if strings.Contains(got, "\r") {
		t.Fatalf("output still contains CR: %q", got)
	}
}

// TestSemicolonSeparators pins the treatment of ';' statement separators.
// The separator is load-bearing on the assembly path, where the parser reads
// semicolon-separated statements: a formatter that drops it fuses two
// statements into a line the assembler rejects, which is data corruption.
// Each row pins the canonical spelling, one space after the ';', tight
// before it, the way the toolchain's own sources and listings write it.
func TestSemicolonSeparators(t *testing.T) {
	cases := []struct {
		name string
		in   string
		want string
	}{
		{
			name: "between instructions, tight",
			in:   "TEXT ·f(SB), $0\nBYTE $0x48;BYTE $0xc7\nRET\n",
			want: "TEXT ·f(SB), $0\n\tBYTE $0x48; BYTE $0xc7\n\tRET\n",
		},
		{
			name: "between instructions, spaced",
			in:   "TEXT ·f(SB), $0\nBYTE $0x48 ; BYTE $0xc7\nRET\n",
			want: "TEXT ·f(SB), $0\n\tBYTE $0x48; BYTE $0xc7\n\tRET\n",
		},
		{
			name: "after a label",
			in:   "TEXT ·f(SB), $0\nlabel: BYTE $1; BYTE $2\nRET\n",
			want: "TEXT ·f(SB), $0\nlabel:\n\tBYTE $1; BYTE $2\n\tRET\n",
		},
		{
			// The continuation-spliced macro shape of the runtime sources:
			// the lexer makes one logical line of the backslash continuations.
			name: "inside a macro body, continued",
			in:   "#define MOVLTOREG(v, off) \\\n\tMOVL $v, AX; \\\n\tMOVL AX, ret+off(FP)\n",
			want: "#define MOVLTOREG(v, off) MOVL $v, AX; MOVL AX, ret+off(FP)\n",
		},
		{
			name: "inside a macro body, one line",
			in:   "#define PEAS BYTE $0x0a; BYTE $0x0b\n",
			want: "#define PEAS BYTE $0x0a; BYTE $0x0b\n",
		},
		{
			name: "several separators in one line",
			in:   "TEXT ·f(SB), $0\nBYTE $1; BYTE $2; BYTE $3\nRET\n",
			want: "TEXT ·f(SB), $0\n\tBYTE $1; BYTE $2; BYTE $3\n\tRET\n",
		},
		{
			name: "two separators back to back",
			in:   "TEXT ·f(SB), $0\nBYTE $1;; BYTE $2\nRET\n",
			want: "TEXT ·f(SB), $0\n\tBYTE $1;; BYTE $2\n\tRET\n",
		},
		{
			name: "inside a line comment, untouched",
			in:   "TEXT ·f(SB), $0\n// keep; the; separators\nBYTE $1\nRET\n",
			want: "TEXT ·f(SB), $0\n\t// keep; the; separators\n\tBYTE $1\n\tRET\n",
		},
		{
			name: "after a statement, before a comment",
			in:   "TEXT ·f(SB), $0\nMOVQ AX, BX; // tail\nRET\n",
			want: "TEXT ·f(SB), $0\n\tMOVQ AX, BX; // tail\n\tRET\n",
		},
		{
			name: "last character on a line",
			in:   "TEXT ·f(SB), $0\nBYTE $1;\nRET\n",
			want: "TEXT ·f(SB), $0\n\tBYTE $1;\n\tRET\n",
		},
	}
	for _, tc := range cases {
		t.Run(tc.name, func(t *testing.T) {
			got := Source(tc.in)
			if got != tc.want {
				t.Fatalf("formatting mismatch:\n--- got ---\n%q\n--- want ---\n%q", got, tc.want)
			}
			if again := Source(got); again != got {
				t.Fatalf("not idempotent:\n%q", again)
			}
			if in, out := strings.Count(tc.in, ";"), strings.Count(got, ";"); in != out {
				t.Fatalf("semicolon count changed: %d -> %d\n%s", in, out, got)
			}
			if _, errs := parser.Parse("in.s", got); len(errs) > 0 {
				t.Fatalf("formatted output no longer parses: %v", errs)
			}
		})
	}
}

// TestSemicolonStatementRoundTrip proves the formatter's contract on the
// path where ';' separates statements: parse the source the way the
// assembler does, format it, re-parse the formatted text and compare the
// statement sequence.  Raw operand texts are token-joined, so they are
// insensitive to the whitespace a format pass chooses, and the comparison
// can only fail when a token is lost: dropping a ';' fuses two statements
// into one, exactly the corruption the released formatter committed.
func TestSemicolonStatementRoundTrip(t *testing.T) {
	src := "#define MOVLTOREG(v, off) \\\n" +
		"\tMOVL $v, AX; \\\n" +
		"\tMOVL AX, ret+off(FP)\n" +
		"\n" +
		"TEXT ·f(SB), NOSPLIT, $0\n" +
		"BYTE $0x48; BYTE $0xc7\n" +
		"first: BYTE $1; BYTE $2\n" +
		"MOVLTOREG($42, 0)\n" +
		"RET\n"

	before, errs := parser.ParseWithOptions("in.s", src, parser.Options{Expand: true})
	if len(errs) > 0 {
		t.Fatalf("source does not parse: %v", errs)
	}
	formatted := Source(src)
	after, errs := parser.ParseWithOptions("in.s", formatted, parser.Options{Expand: true})
	if len(errs) > 0 {
		t.Fatalf("formatted source does not parse: %v", errs)
	}

	want, got := stmtSignature(before), stmtSignature(after)
	if !slices.Equal(got, want) {
		t.Fatalf("statement sequence changed:\n--- before ---\n%q\n--- after ---\n%q", want, got)
	}
	if again := Source(formatted); again != formatted {
		t.Fatalf("not idempotent:\n%q", again)
	}

	// The two BYTE statements on the first line must stay two: one fused
	// statement here is the exact defect this package once shipped.
	var bytes []string
	for _, stmt := range stmtSignature(after) {
		if rest, ok := strings.CutPrefix(stmt, "instr BYTE "); ok {
			bytes = append(bytes, rest)
		}
	}
	if want := []string{"$ 0x48", "$ 0xc7", "$ 1", "$ 2"}; !slices.Equal(bytes, want) {
		t.Fatalf("BYTE statements after expansion = %q, want %q", bytes, want)
	}
}

// stmtSignature flattens a parsed file into one string per declaration and
// statement, in source order.  Every component is token-derived, so the
// signature is stable across format passes and moves only when a token is
// lost or gained.
func stmtSignature(f *ast.File) []string {
	var out []string
	for _, d := range f.Decls {
		switch d := d.(type) {
		case *ast.Text:
			out = append(out, "text "+d.Name.Raw)
			for _, s := range d.Body {
				out = append(out, stmtText(s))
			}
		case *ast.Globl:
			out = append(out, "globl "+d.Name.Raw)
		case *ast.Data:
			out = append(out, "data "+d.Name.Raw)
		case *ast.Include:
			out = append(out, "include "+d.Header.Text)
		case *ast.Preproc:
			out = append(out, "preproc "+d.Raw)
		}
	}
	for _, s := range f.Orphans {
		out = append(out, stmtText(s))
	}
	return out
}

// stmtText renders one statement for stmtSignature.
func stmtText(s ast.Stmt) string {
	switch s := s.(type) {
	case *ast.Label:
		return "label " + s.Name.Text
	case *ast.Instr:
		parts := make([]string, 0, len(s.Operands)+1)
		parts = append(parts, s.Mnemonic.Text)
		for _, op := range s.Operands {
			parts = append(parts, op.Raw)
		}
		return "instr " + strings.Join(parts, " ")
	default:
		return "stmt"
	}
}

// lexOperands lexes a single operand string and drops the EOF token.
func lexOperands(s string) []token.Token {
	toks := lexer.Tokenize(s)
	return toks[:len(toks)-1] // drop trailing EOF
}

func TestIdempotent(t *testing.T) {
	src, err := os.ReadFile("../testdata/sample_amd64.s")
	if err != nil {
		t.Fatal(err)
	}
	once := Source(string(src))
	twice := Source(once)
	if once != twice {
		t.Fatal("formatting is not idempotent on the fixture")
	}
}

// TestRoundTrip checks that formatting produces source that still parses
// cleanly on the in-repository fixture.
func TestRoundTrip(t *testing.T) {
	files := []string{"../testdata/sample_amd64.s"}
	for _, path := range files {
		src, err := os.ReadFile(path)
		if err != nil {
			t.Fatal(err)
		}
		formatted := Source(string(src))
		if _, errs := parser.Parse(path, formatted); len(errs) > 0 {
			t.Errorf("formatted %s no longer parses: %v", path, errs)
		}
		if strings.TrimSpace(formatted) == "" {
			t.Errorf("formatted %s is empty", path)
		}
	}
}