os: parse command line without shell32.dll

Go uses CommandLineToArgV from shell32.dll to parse command
line parameters. But shell32.dll is slow to load. Implement
Windows command line parsing in Go. This should make starting
Go programs faster.

I can see these speed ups for runtime.BenchmarkRunningGoProgram

on my Windows 7 amd64:
name                old time/op  new time/op  delta
RunningGoProgram-2  11.2ms ± 1%  10.4ms ± 2%  -6.63%  (p=0.000 n=9+10)

on my Windows XP 386:
name                old time/op  new time/op  delta
RunningGoProgram-2  19.0ms ± 3%  12.1ms ± 1%  -36.20%  (p=0.000 n=10+10)

on @egonelbre Windows 10 amd64:
name                old time/op  new time/op  delta
RunningGoProgram-8  17.0ms ± 1%  15.3ms ± 2%  -9.71%  (p=0.000 n=10+10)

This CL is based on CL 22932 by John Starks.

Fixes #15588.

Change-Id: Ib14be0206544d0d4492ca1f0d91fac968be52241
Reviewed-on: https://go-review.googlesource.com/37915Reviewed-by: Brad Fitzpatrick <bradfitz@golang.org>
Run-TryBot: Brad Fitzpatrick <bradfitz@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>

src/os/exec_windows.go

@@ -97,17 +97,79 @@ func findProcess(pid int) (p *Process, err error) {
 }
 
 func init() {
-	var argc int32
-	cmd := syscall.GetCommandLine()
-	argv, e := syscall.CommandLineToArgv(cmd, &argc)
-	if e != nil {
-		return
+	p := syscall.GetCommandLine()
+	cmd := syscall.UTF16ToString((*[0xffff]uint16)(unsafe.Pointer(p))[:])
+	if len(cmd) == 0 {
+		arg0, _ := Executable()
+		Args = []string{arg0}
+	} else {
+		Args = commandLineToArgv(cmd)
 	}
-	defer syscall.LocalFree(syscall.Handle(uintptr(unsafe.Pointer(argv))))
-	Args = make([]string, argc)
-	for i, v := range (*argv)[:argc] {
-		Args[i] = syscall.UTF16ToString((*v)[:])
+}
+
+// appendBSBytes appends n '\\' bytes to b and returns the resulting slice.
+func appendBSBytes(b []byte, n int) []byte {
+	for ; n > 0; n-- {
+		b = append(b, '\\')
+	}
+	return b
+}
+
+// readNextArg splits command line string cmd into next
+// argument and command line remainder.
+func readNextArg(cmd string) (arg []byte, rest string) {
+	var b []byte
+	var inquote bool
+	var nslash int
+	for ; len(cmd) > 0; cmd = cmd[1:] {
+		c := cmd[0]
+		switch c {
+		case ' ', '\t':
+			if !inquote {
+				return appendBSBytes(b, nslash), cmd[1:]
+			}
+		case '"':
+			b = appendBSBytes(b, nslash/2)
+			if nslash%2 == 0 {
+				// use "Prior to 2008" rule from
+				// http://daviddeley.com/autohotkey/parameters/parameters.htm
+				// section 5.2 to deal with double double quotes
+				if inquote && len(cmd) > 1 && cmd[1] == '"' {
+					b = append(b, c)
+					cmd = cmd[1:]
+				}
+				inquote = !inquote
+			} else {
+				b = append(b, c)
+			}
+			nslash = 0
+			continue
+		case '\\':
+			nslash++
+			continue
+		}
+		b = appendBSBytes(b, nslash)
+		nslash = 0
+		b = append(b, c)
+	}
+	return appendBSBytes(b, nslash), ""
+}
+
+// commandLineToArgv splits a command line into individual argument
+// strings, following the Windows conventions documented
+// at http://daviddeley.com/autohotkey/parameters/parameters.htm#WINARGV
+func commandLineToArgv(cmd string) []string {
+	var args []string
+	for len(cmd) > 0 {
+		if cmd[0] == ' ' || cmd[0] == '\t' {
+			cmd = cmd[1:]
+			continue
+		}
+		var arg []byte
+		arg, cmd = readNextArg(cmd)
+		args = append(args, string(arg))
 	}
+	return args
 }
 
 func ftToDuration(ft *syscall.Filetime) time.Duration {

src/os/export_windows_test.go

@@ -7,6 +7,7 @@ package os
 // Export for testing.
 
 var (
-	FixLongPath    = fixLongPath
-	NewConsoleFile = newConsoleFile
+	FixLongPath       = fixLongPath
+	NewConsoleFile    = newConsoleFile
+	CommandLineToArgv = commandLineToArgv
 )

src/os/os_windows_test.go

@@ -723,3 +723,137 @@ func TestStatPagefile(t *testing.T) {
 	}
 	t.Fatal(err)
 }
+
+// syscallCommandLineToArgv calls syscall.CommandLineToArgv
+// and converts returned result into []string.
+func syscallCommandLineToArgv(cmd string) ([]string, error) {
+	var argc int32
+	argv, err := syscall.CommandLineToArgv(&syscall.StringToUTF16(cmd)[0], &argc)
+	if err != nil {
+		return nil, err
+	}
+	defer syscall.LocalFree(syscall.Handle(uintptr(unsafe.Pointer(argv))))
+
+	var args []string
+	for _, v := range (*argv)[:argc] {
+		args = append(args, syscall.UTF16ToString((*v)[:]))
+	}
+	return args, nil
+}
+
+// compareCommandLineToArgvWithSyscall ensures that
+// os.CommandLineToArgv(cmd) and syscall.CommandLineToArgv(cmd)
+// return the same result.
+func compareCommandLineToArgvWithSyscall(t *testing.T, cmd string) {
+	syscallArgs, err := syscallCommandLineToArgv(cmd)
+	if err != nil {
+		t.Fatal(err)
+	}
+	args := os.CommandLineToArgv(cmd)
+	if want, have := fmt.Sprintf("%q", syscallArgs), fmt.Sprintf("%q", args); want != have {
+		t.Errorf("testing os.commandLineToArgv(%q) failed: have %q want %q", cmd, args, syscallArgs)
+		return
+	}
+}
+
+func TestCmdArgs(t *testing.T) {
+	tmpdir, err := ioutil.TempDir("", "TestCmdArgs")
+	if err != nil {
+		t.Fatal(err)
+	}
+	defer os.RemoveAll(tmpdir)
+
+	const prog = `
+package main
+
+import (
+	"fmt"
+	"os"
+)
+
+func main() {
+	fmt.Printf("%q", os.Args)
+}
+`
+	src := filepath.Join(tmpdir, "main.go")
+	err = ioutil.WriteFile(src, []byte(prog), 0666)
+	if err != nil {
+		t.Fatal(err)
+	}
+
+	exe := filepath.Join(tmpdir, "main.exe")
+	cmd := osexec.Command("go", "build", "-o", exe, src)
+	cmd.Dir = tmpdir
+	out, err := cmd.CombinedOutput()
+	if err != nil {
+		t.Fatalf("building main.exe failed: %v\n%s", err, out)
+	}
+
+	var cmds = []string{
+		``,
+		` a b c`,
+		` "`,
+		` ""`,
+		` """`,
+		` "" a`,
+		` "123"`,
+		` \"123\"`,
+		` \"123 456\"`,
+		` \\"`,
+		` \\\"`,
+		` \\\\\"`,
+		` \\\"x`,
+		` """"\""\\\"`,
+		` abc`,
+		` \\\\\""x"""y z`,
+		"\tb\t\"x\ty\"",
+		` "Брад" d e`,
+		// examples from https://msdn.microsoft.com/en-us/library/17w5ykft.aspx
+		` "abc" d e`,
+		` a\\b d"e f"g h`,
+		` a\\\"b c d`,
+		` a\\\\"b c" d e`,
+		// http://daviddeley.com/autohotkey/parameters/parameters.htm#WINARGV
+		// from 5.4  Examples
+		` CallMeIshmael`,
+		` "Call Me Ishmael"`,
+		` Cal"l Me I"shmael`,
+		` CallMe\"Ishmael`,
+		` "CallMe\"Ishmael"`,
+		` "Call Me Ishmael\\"`,
+		` "CallMe\\\"Ishmael"`,
+		` a\\\b`,
+		` "a\\\b"`,
+		// from 5.5  Some Common Tasks
+		` "\"Call Me Ishmael\""`,
+		` "C:\TEST A\\"`,
+		` "\"C:\TEST A\\\""`,
+		// from 5.6  The Microsoft Examples Explained
+		` "a b c"  d  e`,
+		` "ab\"c"  "\\"  d`,
+		` a\\\b d"e f"g h`,
+		` a\\\"b c d`,
+		` a\\\\"b c" d e`,
+		// from 5.7  Double Double Quote Examples (pre 2008)
+		` "a b c""`,
+		` """CallMeIshmael"""  b  c`,
+		` """Call Me Ishmael"""`,
+		` """"Call Me Ishmael"" b c`,
+	}
+	for _, cmd := range cmds {
+		compareCommandLineToArgvWithSyscall(t, "test"+cmd)
+		compareCommandLineToArgvWithSyscall(t, `"cmd line"`+cmd)
+		compareCommandLineToArgvWithSyscall(t, exe+cmd)
+
+		// test both syscall.EscapeArg and os.commandLineToArgv
+		args := os.CommandLineToArgv(exe + cmd)
+		out, err := osexec.Command(args[0], args[1:]...).CombinedOutput()
+		if err != nil {
+			t.Fatalf("runing %q failed: %v\n%v", args, err, string(out))
+		}
+		if want, have := fmt.Sprintf("%q", args), string(out); want != have {
+			t.Errorf("wrong output of executing %q: have %q want %q", args, have, want)
+			continue
+		}
+	}
+}