cmd/evm, core/asm: add EVM assembler (#3686)

The evm compile command implements a simple assembly language that compiles to
EVM bytecode.

core/asm/compiler.go

@@ -0,0 +1,281 @@
+// Copyright 2017 The go-ethereum Authors
+// This file is part of the go-ethereum library.
+//
+// The go-ethereum library is free software: you can redistribute it and/or modify
+// it under the terms of the GNU Lesser General Public License as published by
+// the Free Software Foundation, either version 3 of the License, or
+// (at your option) any later version.
+//
+// The go-ethereum library is distributed in the hope that it will be useful,
+// but WITHOUT ANY WARRANTY; without even the implied warranty of
+// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
+// GNU Lesser General Public License for more details.
+//
+// You should have received a copy of the GNU Lesser General Public License
+// along with the go-ethereum library. If not, see <http://www.gnu.org/licenses/>.
+
+package asm
+
+import (
+	"errors"
+	"fmt"
+	"math/big"
+	"os"
+	"strings"
+
+	"github.com/ethereum/go-ethereum/common/math"
+	"github.com/ethereum/go-ethereum/core/vm"
+)
+
+// Compiler contains information about the parsed source
+// and holds the tokens for the program.
+type Compiler struct {
+	tokens []token
+	binary []interface{}
+
+	labels map[string]int
+
+	pc, pos int
+
+	debug bool
+}
+
+// newCompiler returns a new allocated compiler.
+func NewCompiler(debug bool) *Compiler {
+	return &Compiler{
+		labels: make(map[string]int),
+		debug:  debug,
+	}
+}
+
+// Feed feeds tokens in to ch and are interpreted by
+// the compiler.
+//
+// feed is the first pass in the compile stage as it
+// collect the used labels in the program and keeps a
+// program counter which is used to determine the locations
+// of the jump dests. The labels can than be used in the
+// second stage to push labels and determine the right
+// position.
+func (c *Compiler) Feed(ch <-chan token) {
+	for i := range ch {
+		switch i.typ {
+		case number:
+			num := math.MustParseBig256(i.text).Bytes()
+			if len(num) == 0 {
+				num = []byte{0}
+			}
+			c.pc += len(num)
+		case stringValue:
+			c.pc += len(i.text) - 2
+		case element:
+			c.pc++
+		case labelDef:
+			c.labels[i.text] = c.pc
+			c.pc++
+		case label:
+			c.pc += 5
+		}
+
+		c.tokens = append(c.tokens, i)
+	}
+	if c.debug {
+		fmt.Fprintln(os.Stderr, "found", len(c.labels), "labels")
+	}
+}
+
+// Compile compiles the current tokens and returns a
+// binary string that can be interpreted by the EVM
+// and an error if it failed.
+//
+// compile is the second stage in the compile phase
+// which compiles the tokens to EVM instructions.
+func (c *Compiler) Compile() (string, []error) {
+	var errors []error
+	// continue looping over the tokens until
+	// the stack has been exhausted.
+	for c.pos < len(c.tokens) {
+		if err := c.compileLine(); err != nil {
+			errors = append(errors, err)
+		}
+	}
+
+	// turn the binary to hex
+	var bin string
+	for _, v := range c.binary {
+		switch v := v.(type) {
+		case vm.OpCode:
+			bin += fmt.Sprintf("%x", []byte{byte(v)})
+		case []byte:
+			bin += fmt.Sprintf("%x", v)
+		}
+	}
+	return bin, errors
+}
+
+// next returns the next token and increments the
+// posititon.
+func (c *Compiler) next() token {
+	token := c.tokens[c.pos]
+	c.pos++
+	return token
+}
+
+// compile line compiles a single line instruction e.g.
+// "push 1", "jump @labal".
+func (c *Compiler) compileLine() error {
+	n := c.next()
+	if n.typ != lineStart {
+		return compileErr(n, n.typ.String(), lineStart.String())
+	}
+
+	lvalue := c.next()
+	switch lvalue.typ {
+	case eof:
+		return nil
+	case element:
+		if err := c.compileElement(lvalue); err != nil {
+			return err
+		}
+	case labelDef:
+		c.compileLabel()
+	case lineEnd:
+		return nil
+	default:
+		return compileErr(lvalue, lvalue.text, fmt.Sprintf("%v or %v", labelDef, element))
+	}
+
+	if n := c.next(); n.typ != lineEnd {
+		return compileErr(n, n.text, lineEnd.String())
+	}
+
+	return nil
+}
+
+// compileNumber compiles the number to bytes
+func (c *Compiler) compileNumber(element token) (int, error) {
+	num := math.MustParseBig256(element.text).Bytes()
+	if len(num) == 0 {
+		num = []byte{0}
+	}
+	c.pushBin(num)
+	return len(num), nil
+}
+
+// compileElement compiles the element (push & label or both)
+// to a binary representation and may error if incorrect statements
+// where fed.
+func (c *Compiler) compileElement(element token) error {
+	// check for a jump. jumps must be read and compiled
+	// from right to left.
+	if isJump(element.text) {
+		rvalue := c.next()
+		switch rvalue.typ {
+		case number:
+			// TODO figure out how to return the error properly
+			c.compileNumber(rvalue)
+		case stringValue:
+			// strings are quoted, remove them.
+			c.pushBin(rvalue.text[1 : len(rvalue.text)-2])
+		case label:
+			c.pushBin(vm.PUSH4)
+			pos := big.NewInt(int64(c.labels[rvalue.text])).Bytes()
+			pos = append(make([]byte, 4-len(pos)), pos...)
+			c.pushBin(pos)
+		default:
+			return compileErr(rvalue, rvalue.text, "number, string or label")
+		}
+		// push the operation
+		c.pushBin(toBinary(element.text))
+		return nil
+	} else if isPush(element.text) {
+		// handle pushes. pushes are read from left to right.
+		var value []byte
+
+		rvalue := c.next()
+		switch rvalue.typ {
+		case number:
+			value = math.MustParseBig256(rvalue.text).Bytes()
+			if len(value) == 0 {
+				value = []byte{0}
+			}
+		case stringValue:
+			value = []byte(rvalue.text[1 : len(rvalue.text)-1])
+		case label:
+			value = make([]byte, 4)
+			copy(value, big.NewInt(int64(c.labels[rvalue.text])).Bytes())
+		default:
+			return compileErr(rvalue, rvalue.text, "number, string or label")
+		}
+
+		if len(value) > 32 {
+			return fmt.Errorf("%d type error: unsupported string or number with size > 32", rvalue.lineno)
+		}
+
+		c.pushBin(vm.OpCode(int(vm.PUSH1) - 1 + len(value)))
+		c.pushBin(value)
+	} else {
+		c.pushBin(toBinary(element.text))
+	}
+
+	return nil
+}
+
+// compileLabel pushes a jumpdest to the binary slice.
+func (c *Compiler) compileLabel() {
+	c.pushBin(vm.JUMPDEST)
+}
+
+// pushBin pushes the value v to the binary stack.
+func (c *Compiler) pushBin(v interface{}) {
+	if c.debug {
+		fmt.Printf("%d: %v\n", len(c.binary), v)
+	}
+	c.binary = append(c.binary, v)
+}
+
+// isPush returns whether the string op is either any of
+// push(N).
+func isPush(op string) bool {
+	if op == "push" {
+		return true
+	}
+	return false
+}
+
+// isJump returns whether the string op is jump(i)
+func isJump(op string) bool {
+	return op == "jumpi" || op == "jump"
+}
+
+// toBinary converts text to a vm.OpCode
+func toBinary(text string) vm.OpCode {
+	if isPush(text) {
+		text = "push1"
+	}
+	return vm.StringToOp(strings.ToUpper(text))
+}
+
+type compileError struct {
+	got  string
+	want string
+
+	lineno int
+}
+
+func (err compileError) Error() string {
+	return fmt.Sprintf("%d syntax error: unexpected %v, expected %v", err.lineno, err.got, err.want)
+}
+
+var (
+	errExpBol            = errors.New("expected beginning of line")
+	errExpElementOrLabel = errors.New("expected beginning of line")
+)
+
+func compileErr(c token, got, want string) error {
+	return compileError{
+		got:    got,
+		want:   want,
+		lineno: c.lineno,
+	}
+}