Merge pull request #287 from Makpoc/parsewincmd

Fix windows command parsing

middleware/commands.go

@@ -2,18 +2,30 @@ package middleware
 
 import (
 	"errors"
+	"runtime"
+	"unicode"
 
 	"github.com/flynn/go-shlex"
 )
 
+var runtimeGoos = runtime.GOOS
+
 // SplitCommandAndArgs takes a command string and parses it
 // shell-style into the command and its separate arguments.
 func SplitCommandAndArgs(command string) (cmd string, args []string, err error) {
-	parts, err := shlex.Split(command)
+	var parts []string
+
+	if runtimeGoos == "windows" {
+		parts = parseWindowsCommand(command) // parse it Windows-style
+	} else {
+		parts, err = parseUnixCommand(command) // parse it Unix-style
 		if err != nil {
 			err = errors.New("error parsing command: " + err.Error())
 			return
-	} else if len(parts) == 0 {
+		}
+	}
+
+	if len(parts) == 0 {
 		err = errors.New("no command contained in '" + command + "'")
 		return
 	}
@@ -25,3 +37,84 @@ func SplitCommandAndArgs(command string) (cmd string, args []string, err error)
 
 	return
 }
+
+// parseUnixCommand parses a unix style command line and returns the
+// command and its arguments or an error
+func parseUnixCommand(cmd string) ([]string, error) {
+	return shlex.Split(cmd)
+}
+
+// parseWindowsCommand parses windows command lines and
+// returns the command and the arguments as an array. It
+// should be able to parse commonly used command lines.
+// Only basic syntax is supported:
+//  - spaces in double quotes are not token delimiters
+//  - double quotes are escaped by either backspace or another double quote
+//  - except for the above case backspaces are path separators (not special)
+//
+// Many sources point out that escaping quotes using backslash can be unsafe.
+// Use two double quotes when possible. (Source: http://stackoverflow.com/a/31413730/2616179 )
+//
+// This function has to be used on Windows instead
+// of the shlex package because this function treats backslash
+// characters properly.
+func parseWindowsCommand(cmd string) []string {
+	const backslash = '\\'
+	const quote = '"'
+
+	var parts []string
+	var part string
+	var inQuotes bool
+	var lastRune rune
+
+	for i, ch := range cmd {
+
+		if i != 0 {
+			lastRune = rune(cmd[i-1])
+		}
+
+		if ch == backslash {
+			// put it in the part - for now we don't know if it's an
+			// escaping char or path separator
+			part += string(ch)
+			continue
+		}
+
+		if ch == quote {
+			if lastRune == backslash {
+				// remove the backslash from the part and add the escaped quote instead
+				part = part[:len(part)-1]
+				part += string(ch)
+				continue
+			}
+
+			if lastRune == quote {
+				// revert the last change of the inQuotes state
+				// it was an escaping quote
+				inQuotes = !inQuotes
+				part += string(ch)
+				continue
+			}
+
+			// normal escaping quotes
+			inQuotes = !inQuotes
+			continue
+
+		}
+
+		if unicode.IsSpace(ch) && !inQuotes && len(part) > 0 {
+			parts = append(parts, part)
+			part = ""
+			continue
+		}
+
+		part += string(ch)
+	}
+
+	if len(part) > 0 {
+		parts = append(parts, part)
+		part = ""
+	}
+
+	return parts
+}

middleware/commands_test.go

@@ -2,11 +2,176 @@ package middleware
 
 import (
 	"fmt"
+	"runtime"
 	"strings"
 	"testing"
 )
 
+func TestParseUnixCommand(t *testing.T) {
+	tests := []struct {
+		input    string
+		expected []string
+	}{
+		// 0 - emtpy command
+		{
+			input:    ``,
+			expected: []string{},
+		},
+		// 1 - command without arguments
+		{
+			input:    `command`,
+			expected: []string{`command`},
+		},
+		// 2 - command with single argument
+		{
+			input:    `command arg1`,
+			expected: []string{`command`, `arg1`},
+		},
+		// 3 - command with multiple arguments
+		{
+			input:    `command arg1 arg2`,
+			expected: []string{`command`, `arg1`, `arg2`},
+		},
+		// 4 - command with single argument with space character - in quotes
+		{
+			input:    `command "arg1 arg1"`,
+			expected: []string{`command`, `arg1 arg1`},
+		},
+		// 5 - command with multiple spaces and tab character
+		{
+			input:    "command arg1    arg2\targ3",
+			expected: []string{`command`, `arg1`, `arg2`, `arg3`},
+		},
+		// 6 - command with single argument with space character - escaped with backspace
+		{
+			input:    `command arg1\ arg2`,
+			expected: []string{`command`, `arg1 arg2`},
+		},
+		// 7 - single quotes should escape special chars
+		{
+			input:    `command 'arg1\ arg2'`,
+			expected: []string{`command`, `arg1\ arg2`},
+		},
+	}
+
+	for i, test := range tests {
+		errorPrefix := fmt.Sprintf("Test [%d]: ", i)
+		errorSuffix := fmt.Sprintf(" Command to parse: [%s]", test.input)
+		actual, _ := parseUnixCommand(test.input)
+		if len(actual) != len(test.expected) {
+			t.Errorf(errorPrefix+"Expected %d parts, got %d: %#v."+errorSuffix, len(test.expected), len(actual), actual)
+			continue
+		}
+		for j := 0; j < len(actual); j++ {
+			if expectedPart, actualPart := test.expected[j], actual[j]; expectedPart != actualPart {
+				t.Errorf(errorPrefix+"Expected: %v Actual: %v (index %d)."+errorSuffix, expectedPart, actualPart, j)
+			}
+		}
+	}
+}
+
+func TestParseWindowsCommand(t *testing.T) {
+	tests := []struct {
+		input    string
+		expected []string
+	}{
+		{ // 0 - empty command - do not fail
+			input:    ``,
+			expected: []string{},
+		},
+		{ // 1 - cmd without args
+			input:    `cmd`,
+			expected: []string{`cmd`},
+		},
+		{ // 2 - multiple args
+			input:    `cmd arg1 arg2`,
+			expected: []string{`cmd`, `arg1`, `arg2`},
+		},
+		{ // 3 - multiple args with space
+			input:    `cmd "combined arg" arg2`,
+			expected: []string{`cmd`, `combined arg`, `arg2`},
+		},
+		{ // 4 - path without spaces
+			input:    `mkdir C:\Windows\foo\bar`,
+			expected: []string{`mkdir`, `C:\Windows\foo\bar`},
+		},
+		{ // 5 - command with space in quotes
+			input:    `"command here"`,
+			expected: []string{`command here`},
+		},
+		{ // 6 - argument with escaped quotes (two quotes)
+			input:    `cmd ""arg""`,
+			expected: []string{`cmd`, `"arg"`},
+		},
+		{ // 7 - argument with escaped quotes (backslash)
+			input:    `cmd \"arg\"`,
+			expected: []string{`cmd`, `"arg"`},
+		},
+		{ // 8 - two quotes (escaped) inside an inQuote element
+			input:    `cmd "a ""quoted value"`,
+			expected: []string{`cmd`, `a "quoted value`},
+		},
+		// TODO - see how many quotes are dislayed if we use "", """, """""""
+		{ // 9 - two quotes outside an inQuote element
+			input:    `cmd a ""quoted value`,
+			expected: []string{`cmd`, `a`, `"quoted`, `value`},
+		},
+		{ // 10 - path with space in quotes
+			input:    `mkdir "C:\directory name\foobar"`,
+			expected: []string{`mkdir`, `C:\directory name\foobar`},
+		},
+		{ // 11 - space without quotes
+			input:    `mkdir C:\ space`,
+			expected: []string{`mkdir`, `C:\`, `space`},
+		},
+		{ // 12 - space in quotes
+			input:    `mkdir "C:\ space"`,
+			expected: []string{`mkdir`, `C:\ space`},
+		},
+		{ // 13 - UNC
+			input:    `mkdir \\?\C:\Users`,
+			expected: []string{`mkdir`, `\\?\C:\Users`},
+		},
+		{ // 14 - UNC with space
+			input:    `mkdir "\\?\C:\Program Files"`,
+			expected: []string{`mkdir`, `\\?\C:\Program Files`},
+		},
+
+		{ // 15 - unclosed quotes - treat as if the path ends with quote
+			input:    `mkdir "c:\Program files`,
+			expected: []string{`mkdir`, `c:\Program files`},
+		},
+		{ // 16 - quotes used inside the argument
+			input:    `mkdir "c:\P"rogra"m f"iles`,
+			expected: []string{`mkdir`, `c:\Program files`},
+		},
+	}
+
+	for i, test := range tests {
+		errorPrefix := fmt.Sprintf("Test [%d]: ", i)
+		errorSuffix := fmt.Sprintf(" Command to parse: [%s]", test.input)
+
+		actual := parseWindowsCommand(test.input)
+		if len(actual) != len(test.expected) {
+			t.Errorf(errorPrefix+"Expected %d parts, got %d: %#v."+errorSuffix, len(test.expected), len(actual), actual)
+			continue
+		}
+		for j := 0; j < len(actual); j++ {
+			if expectedPart, actualPart := test.expected[j], actual[j]; expectedPart != actualPart {
+				t.Errorf(errorPrefix+"Expected: %v Actual: %v (index %d)."+errorSuffix, expectedPart, actualPart, j)
+			}
+		}
+	}
+}
+
 func TestSplitCommandAndArgs(t *testing.T) {
+
+	// force linux parsing. It's more robust and covers error cases
+	runtimeGoos = "linux"
+	defer func() {
+		runtimeGoos = runtime.GOOS
+	}()
+
 	var parseErrorContent = "error parsing command:"
 	var noCommandErrContent = "no command contained in"
 
@@ -16,84 +181,42 @@ func TestSplitCommandAndArgs(t *testing.T) {
 		expectedArgs       []string
 		expectedErrContent string
 	}{
-		// Test case 0 - emtpy command
+		// 0 - emtpy command
 		{
 			input:              ``,
 			expectedCommand:    ``,
 			expectedArgs:       nil,
 			expectedErrContent: noCommandErrContent,
 		},
-		// Test case 1 - command without arguments
+		// 1 - command without arguments
 		{
 			input:              `command`,
 			expectedCommand:    `command`,
 			expectedArgs:       nil,
 			expectedErrContent: ``,
 		},
-		// Test case 2 - command with single argument
+		// 2 - command with single argument
 		{
 			input:              `command arg1`,
 			expectedCommand:    `command`,
 			expectedArgs:       []string{`arg1`},
 			expectedErrContent: ``,
 		},
-		// Test case 3 - command with multiple arguments
+		// 3 - command with multiple arguments
 		{
 			input:              `command arg1 arg2`,
 			expectedCommand:    `command`,
 			expectedArgs:       []string{`arg1`, `arg2`},
 			expectedErrContent: ``,
 		},
-		// Test case 4 - command with single argument with space character - in quotes
-		{
-			input:              `command "arg1 arg1"`,
-			expectedCommand:    `command`,
-			expectedArgs:       []string{`arg1 arg1`},
-			expectedErrContent: ``,
-		},
-		// Test case 4 - command with single argument with space character - escaped
-		{
-			input:              `command arg1\ arg1`,
-			expectedCommand:    `command`,
-			expectedArgs:       []string{`arg1 arg1`},
-			expectedErrContent: ``,
-		},
-		// Test case 6 - command with escaped quote character
-		{
-			input:              `command "arg1 \" arg1"`,
-			expectedCommand:    `command`,
-			expectedArgs:       []string{`arg1 " arg1`},
-			expectedErrContent: ``,
-		},
-		// Test case 7 - command with escaped backslash
-		{
-			input:              `command '\arg1'`,
-			expectedCommand:    `command`,
-			expectedArgs:       []string{`\arg1`},
-			expectedErrContent: ``,
-		},
-		// Test case 8 - command with comments
-		{
-			input:              `command arg1 #comment1 comment2`,
-			expectedCommand:    `command`,
-			expectedArgs:       []string{`arg1`},
-			expectedErrContent: "",
-		},
-		// Test case 9 - command with multiple spaces and tab character
-		{
-			input:              "command arg1    arg2\targ3",
-			expectedCommand:    `command`,
-			expectedArgs:       []string{`arg1`, `arg2`, "arg3"},
-			expectedErrContent: "",
-		},
-		// Test case 10 - command with unclosed quotes
+		// 4 - command with unclosed quotes
 		{
 			input:              `command "arg1 arg2`,
 			expectedCommand:    "",
 			expectedArgs:       nil,
 			expectedErrContent: parseErrorContent,
 		},
-		// Test case 11 - command with unclosed quotes
+		// 5 - command with unclosed quotes
 		{
 			input:              `command 'arg1 arg2"`,
 			expectedCommand:    "",
@@ -120,14 +243,14 @@ func TestSplitCommandAndArgs(t *testing.T) {
 
 		// test if command matches
 		if test.expectedCommand != actualCommand {
-			t.Errorf("Expected command: [%s], actual: [%s]."+errorSuffix, test.expectedCommand, actualCommand)
+			t.Errorf(errorPrefix+"Expected command: [%s], actual: [%s]."+errorSuffix, test.expectedCommand, actualCommand)
 		}
 
 		// test if arguments match
 		if len(test.expectedArgs) != len(actualArgs) {
-			t.Errorf("Wrong number of arguments! Expected [%v], actual [%v]."+errorSuffix, test.expectedArgs, actualArgs)
-		}
-
+			t.Errorf(errorPrefix+"Wrong number of arguments! Expected [%v], actual [%v]."+errorSuffix, test.expectedArgs, actualArgs)
+		} else {
+			// test args only if the count matches.
 			for j, actualArg := range actualArgs {
 				expectedArg := test.expectedArgs[j]
 				if actualArg != expectedArg {
@@ -135,4 +258,34 @@ func TestSplitCommandAndArgs(t *testing.T) {
 				}
 			}
 		}
+	}
+}
+
+func ExampleSplitCommandAndArgs() {
+	var commandLine string
+	var command string
+	var args []string
+
+	// just for the test - change GOOS and reset it at the end of the test
+	runtimeGoos = "windows"
+	defer func() {
+		runtimeGoos = runtime.GOOS
+	}()
+
+	commandLine = `mkdir /P "C:\Program Files"`
+	command, args, _ = SplitCommandAndArgs(commandLine)
+
+	fmt.Printf("Windows: %s: %s [%s]\n", commandLine, command, strings.Join(args, ","))
+
+	// set GOOS to linux
+	runtimeGoos = "linux"
+
+	commandLine = `mkdir -p /path/with\ space`
+	command, args, _ = SplitCommandAndArgs(commandLine)
+
+	fmt.Printf("Linux: %s: %s [%s]\n", commandLine, command, strings.Join(args, ","))
+
+	// Output:
+	// Windows: mkdir /P "C:\Program Files": mkdir [/P,C:\Program Files]
+	// Linux: mkdir -p /path/with\ space: mkdir [-p,/path/with space]
 }