add skeleton for instruction constructors

internal/cmd/avogen/main.go

@@ -13,9 +13,10 @@ import (
 )
 
 var generators = map[string]gen.Builder{
-	"asmtest":    gen.NewAsmTest,
+	"asmtest":      gen.NewAsmTest,
-	"godata":     gen.NewGoData,
+	"godata":       gen.NewGoData,
-	"godatatest": gen.NewGoDataTest,
+	"godatatest":   gen.NewGoDataTest,
+	"constructors": gen.NewConstructors,
 }
 
 // Command-line flags.

internal/gen/constructors.go

@@ -0,0 +1,121 @@
+package gen
+
+import (
+	"bytes"
+	"fmt"
+	"sort"
+	"strconv"
+	"strings"
+	"text/tabwriter"
+
+	"github.com/mmcloughlin/avo/internal/inst"
+)
+
+type constructors struct {
+	cfg Config
+	printer
+}
+
+func NewConstructors(cfg Config) Interface {
+	return GoFmt(&constructors{cfg: cfg})
+}
+
+func (c *constructors) Generate(is []inst.Instruction) ([]byte, error) {
+	c.Printf("// %s\n\n", c.cfg.GeneratedWarning())
+	c.Printf("package x86\n\n")
+	c.Printf("import \"github.com/mmcloughlin/avo\"\n\n")
+
+	for _, i := range is {
+		c.instruction(i)
+	}
+
+	return c.Result()
+}
+
+func (c *constructors) instruction(i inst.Instruction) {
+	for _, line := range c.doc(i) {
+		c.Printf("// %s\n", line)
+	}
+
+	paramlist, _ := params(i)
+
+	c.Printf("func %s(%s) error {\n", i.Opcode, paramlist)
+	c.Printf("\treturn nil\n")
+	c.Printf("}\n\n")
+}
+
+// doc generates the lines of the function comment.
+func (c *constructors) doc(i inst.Instruction) []string {
+	lines := []string{
+		fmt.Sprintf("%s: %s.", i.Opcode, i.Summary),
+		"",
+		"Forms:",
+		"",
+	}
+
+	// Write a table of instruction forms.
+	buf := bytes.NewBuffer(nil)
+	w := tabwriter.NewWriter(buf, 0, 0, 1, ' ', 0)
+	for _, f := range i.Forms {
+		row := i.Opcode + "\t" + strings.Join(f.Signature(), "\t") + "\n"
+		fmt.Fprint(w, row)
+	}
+	w.Flush()
+
+	tbl := strings.TrimSpace(buf.String())
+	for _, line := range strings.Split(tbl, "\n") {
+		lines = append(lines, "\t"+line)
+	}
+
+	return lines
+}
+
+// params generates the function parameters and a function.
+func params(i inst.Instruction) (string, func(int) string) {
+	a := i.Arities()
+
+	// Handle the case of forms with multiple arities.
+	if len(a) > 1 {
+		return "ops ...avo.Operand", func(j int) string {
+			return fmt.Sprintf("ops[%d]", j)
+		}
+	}
+
+	// All forms have the same arity.
+	n := a[0]
+	if n == 0 {
+		return "", func(int) string { panic("unreachable") }
+	}
+
+	// Generate nice-looking variable names.
+	ops := make([]string, n)
+	count := map[string]int{}
+	for j := 0; j < n; j++ {
+		// Collect unique lowercase bytes from first characters of operand types.
+		s := map[byte]bool{}
+		for _, f := range i.Forms {
+			c := f.Operands[j].Type[0]
+			if 'a' <= c && c <= 'z' {
+				s[c] = true
+			}
+		}
+
+		// Operand name is the sorted bytes.
+		var b []byte
+		for c := range s {
+			b = append(b, c)
+		}
+		sort.Slice(b, func(i, j int) bool { return b[i] < b[j] })
+		name := string(b)
+
+		// Append a counter if we've seen it already.
+		m := count[name]
+		count[name]++
+		if m > 0 {
+			name += strconv.Itoa(m)
+		}
+		ops[j] = name
+	}
+
+	return strings.Join(ops, ", ") + " avo.Operand", func(j int) string { return ops[j] }
+}

internal/gen/constructors_test.go

@@ -0,0 +1,28 @@
+package gen
+
+import (
+	"testing"
+
+	"github.com/mmcloughlin/avo/internal/inst"
+)
+
+func TestParamsUniqueArgNames(t *testing.T) {
+	for _, i := range inst.Instructions {
+		_, argname := params(i)
+		for _, n := range i.Arities() {
+			if n == 0 {
+				continue
+			}
+			names := map[string]bool{}
+			for j := 0; j < n; j++ {
+				names[argname(j)] = true
+			}
+			if len(names) != n {
+				t.Errorf("repeated argument for instruction %s", i.Opcode)
+			}
+			if _, found := names[""]; found {
+				t.Errorf("empty argument name for instruction %s", i.Opcode)
+			}
+		}
+	}
+}

internal/gen/gen_test.go

@@ -7,5 +7,6 @@ func TestBuilderInterfaces(t *testing.T) {
 		NewAsmTest,
 		NewGoData,
 		NewGoDataTest,
+		NewConstructors,
 	}
 }

internal/inst/table_test.go

@@ -40,6 +40,9 @@ func TestInstructionProperties(t *testing.T) {
 		if len(i.Forms) == 0 {
 			t.Errorf("instruction %s has no forms", i.Opcode)
 		}
+		if len(i.Arities()) == 0 {
+			t.Errorf("instruction %s has empty arities list", i.Opcode)
+		}
 	}
 }
 

internal/inst/types.go

@@ -28,6 +28,14 @@ type Form struct {
 	ImplicitOperands []ImplicitOperand
 }
 
+func (f Form) Signature() []string {
+	s := make([]string, len(f.Operands))
+	for i, op := range f.Operands {
+		s[i] = op.Type
+	}
+	return s
+}
+
 type Operand struct {
 	Type   string
 	Action Action

x86/make.go

@@ -1,74 +0,0 @@
-// +build ignore
-
-package main
-
-import (
-	"flag"
-	"fmt"
-	"go/build"
-	"log"
-	"os"
-	"path/filepath"
-	"regexp"
-	"sort"
-
-	"golang.org/x/arch/x86/x86csv"
-)
-
-const csvname = "x86.v0.2.csv"
-
-var csvpath = flag.String(
-	"csv",
-	filepath.Join(build.Default.GOPATH, "src/golang.org/x/arch/x86", csvname),
-	"path to "+csvname,
-)
-
-func UniqueOpCodes(instrs []*x86csv.Inst) []string {
-	set := map[string]bool{}
-	for _, i := range instrs {
-		set[i.GoOpcode()] = true
-	}
-
-	opcodes := make([]string, 0, len(set))
-	for opcode := range set {
-		opcodes = append(opcodes, opcode)
-	}
-
-	sort.Strings(opcodes)
-
-	return opcodes
-}
-
-var functionRegex = regexp.MustCompile(`^[A-Z0-9]+$`)
-
-func main() {
-	flag.Parse()
-
-	f, err := os.Open(*csvpath)
-	if err != nil {
-		log.Fatal(err)
-	}
-	defer f.Close()
-
-	r := x86csv.NewReader(f)
-
-	instrs, err := r.ReadAll()
-	if err != nil {
-		log.Fatal(err)
-	}
-
-	fmt.Print("package x86\n\nimport \"github.com/mmcloughlin/avo\"\n")
-
-	for _, opcode := range UniqueOpCodes(instrs) {
-		if !functionRegex.MatchString(opcode) {
-			log.Printf("skip %s", opcode)
-			continue
-		}
-		fmt.Printf("\nfunc %s(f *avo.Function, operands ...string) {\n", opcode)
-		fmt.Printf("\tf.AddInstruction(avo.Instruction{\n")
-		fmt.Printf("\t\tMnemonic: \"%s\",\n", opcode)
-		fmt.Printf("\t\tOperands: operands,\n")
-		fmt.Printf("\t})\n")
-		fmt.Printf("}\n")
-	}
-}