bash-for-windows/internal/shell/expand.go

package shell

import (
	"bytes"
	"fmt"
	"os"
	"path/filepath"
	"strconv"
	"strings"
)

func isVarChar(c byte) bool {
	return (c >= 'a' && c <= 'z') || (c >= 'A' && c <= 'Z') || (c >= '0' && c <= '9') || c == '_'
}

// expandWord expands $VAR, ${VAR}, $(...), $((...)) in a single token.
// Quote characters (single/double) are interpreted here and stripped from output.
func (s *Shell) expandWord(word string) string {
	// Tilde expansion (only when not quoted)
	if strings.HasPrefix(word, "~") {
		home := s.GetVar("HOME")
		if home == "" {
			home = os.Getenv("USERPROFILE")
		}
		if len(word) == 1 {
			return home
		}
		if word[1] == '/' || word[1] == '\\' {
			return home + word[1:]
		}
	}

	var result strings.Builder
	inSingle := false
	inDouble := false
	i := 0

	for i < len(word) {
		ch := word[i]

		switch {
		case ch == '\'' && !inDouble:
			inSingle = !inSingle
			i++
		case ch == '"' && !inSingle:
			inDouble = !inDouble
			i++
		case ch == '\\' && !inSingle:
			if i+1 < len(word) {
				next := word[i+1]
				if inDouble {
					// In double quotes, only certain chars are escaped
					switch next {
					case '$', '`', '"', '\\', '\n':
						result.WriteByte(next)
						i += 2
					default:
						result.WriteByte('\\')
						result.WriteByte(next)
						i += 2
					}
				} else {
					result.WriteByte(next)
					i += 2
				}
			} else {
				i++
			}
		case ch == '$' && !inSingle:
			i++ // skip $
			if i >= len(word) {
				result.WriteByte('$')
				break
			}
			switch word[i] {
			case '(':
				if i+1 < len(word) && word[i+1] == '(' {
					// $(( arithmetic ))
					j := i + 2
					depth := 2
					for j < len(word) {
						if word[j] == '(' {
							depth++
						}
						if word[j] == ')' {
							depth--
							if depth == 0 {
								j++
								break
							}
						}
						j++
					}
					expr := word[i+2 : j-2]
					result.WriteString(strconv.Itoa(s.evalArith(expr)))
					i = j
				} else {
					// $( command substitution )
					j := i + 1
					depth := 1
					for j < len(word) {
						if word[j] == '(' {
							depth++
						}
						if word[j] == ')' {
							depth--
							if depth == 0 {
								break
							}
						}
						j++
					}
					cmd := word[i+1 : j]
					out := s.captureCommand(cmd)
					result.WriteString(strings.TrimRight(out, "\n"))
					i = j + 1
				}
			case '{':
				j := i + 1
				depth := 1
				for j < len(word) {
					if word[j] == '{' {
						depth++
					}
					if word[j] == '}' {
						depth--
						if depth == 0 {
							break
						}
					}
					j++
				}
				varExpr := word[i+1 : j]
				result.WriteString(s.evalVarExpr(varExpr))
				i = j + 1
			case '?':
				result.WriteString(s.vars["?"])
				i++
			case '$':
				result.WriteString(fmt.Sprintf("%d", os.Getpid()))
				i++
			case '!':
				result.WriteString(s.vars["!"])
				i++
			case '#':
				result.WriteString(s.vars["#"])
				i++
			case '@':
				result.WriteString(s.vars["@"])
				i++
			case '*':
				result.WriteString(s.vars["*"])
				i++
			default:
				j := i
				for j < len(word) && isVarChar(word[j]) {
					j++
				}
				if j == i {
					result.WriteByte('$')
				} else {
					result.WriteString(s.getVar(word[i:j]))
					i = j
				}
			}
		default:
			result.WriteByte(ch)
			i++
		}
	}
	return result.String()
}

func (s *Shell) getVar(name string) string {
	if v, ok := s.vars[name]; ok {
		return v
	}
	return os.Getenv(name)
}

func (s *Shell) evalVarExpr(expr string) string {
	// ${#VAR} — string length
	if strings.HasPrefix(expr, "#") {
		return strconv.Itoa(len(s.getVar(expr[1:])))
	}
	// ${VAR:-default}
	if idx := strings.Index(expr, ":-"); idx >= 0 {
		varName := expr[:idx]
		if v := s.getVar(varName); v != "" {
			return v
		}
		return s.expandWord(expr[idx+2:])
	}
	// ${VAR:=default}
	if idx := strings.Index(expr, ":="); idx >= 0 {
		varName := expr[:idx]
		if v := s.getVar(varName); v != "" {
			return v
		}
		expanded := s.expandWord(expr[idx+2:])
		s.vars[varName] = expanded
		return expanded
	}
	// ${VAR:+alt}
	if idx := strings.Index(expr, ":+"); idx >= 0 {
		varName := expr[:idx]
		if v := s.getVar(varName); v != "" {
			return s.expandWord(expr[idx+2:])
		}
		return ""
	}
	// ${VAR%pattern} — strip shortest suffix
	if idx := strings.Index(expr, "%"); idx >= 0 {
		varName := expr[:idx]
		pattern := expr[idx+1:]
		v := s.getVar(varName)
		if strings.HasSuffix(v, pattern) {
			return v[:len(v)-len(pattern)]
		}
		return v
	}
	// ${VAR#pattern} — strip shortest prefix
	if idx := strings.Index(expr, "#"); idx >= 0 {
		varName := expr[:idx]
		pattern := expr[idx+1:]
		v := s.getVar(varName)
		if strings.HasPrefix(v, pattern) {
			return v[len(pattern):]
		}
		return v
	}
	return s.getVar(expr)
}

// captureCommand runs a command and returns its stdout as a string.
func (s *Shell) captureCommand(cmd string) string {
	var buf bytes.Buffer
	s.withIO(nil, &buf, nil, func() error {
		return s.Execute(cmd)
	})
	return buf.String()
}

// evalArith evaluates a shell arithmetic expression.
func (s *Shell) evalArith(expr string) int {
	expr = strings.TrimSpace(s.expandWord(expr))
	// Expand bare variable names (e.g. i+1 → value_of_i + 1)
	expr = s.expandArithVars(expr)
	return evalArithExpr(expr)
}

// expandArithVars replaces bare identifier names with their shell variable values.
func (s *Shell) expandArithVars(expr string) string {
	var result strings.Builder
	i := 0
	for i < len(expr) {
		c := expr[i]
		// Identifier start (letter or _), but not a digit
		if (c >= 'a' && c <= 'z') || (c >= 'A' && c <= 'Z') || c == '_' {
			j := i
			for j < len(expr) && isVarChar(expr[j]) {
				j++
			}
			varName := expr[i:j]
			val := s.getVar(varName)
			if val == "" {
				val = "0"
			}
			result.WriteString(val)
			i = j
		} else {
			result.WriteByte(c)
			i++
		}
	}
	return result.String()
}

func evalArithExpr(expr string) int {
	expr = strings.TrimSpace(expr)
	if n, err := strconv.Atoi(expr); err == nil {
		return n
	}
	// Strip outer parens
	if strings.HasPrefix(expr, "(") && strings.HasSuffix(expr, ")") {
		return evalArithExpr(expr[1 : len(expr)-1])
	}
	// Operators in precedence order (lowest first so we split on last occurrence)
	for _, op := range []string{"+", "-", "*", "/", "%"} {
		if idx := findBinaryOp(expr, op); idx >= 0 {
			left := evalArithExpr(expr[:idx])
			right := evalArithExpr(expr[idx+1:])
			switch op {
			case "+":
				return left + right
			case "-":
				return left - right
			case "*":
				return left * right
			case "/":
				if right == 0 {
					return 0
				}
				return left / right
			case "%":
				if right == 0 {
					return 0
				}
				return left % right
			}
		}
	}
	return 0
}

func findBinaryOp(expr, op string) int {
	depth := 0
	for i := len(expr) - 1; i >= 0; i-- {
		switch expr[i] {
		case ')':
			depth++
		case '(':
			depth--
		}
		if depth != 0 {
			continue
		}
		if expr[i:i+1] == op {
			if (op == "-" || op == "+") && i == 0 {
				continue
			}
			return i
		}
	}
	return -1
}

// expandGlob expands glob patterns; returns original if no match.
func (s *Shell) expandGlob(word string) []string {
	if !strings.ContainsAny(word, "*?[") {
		return []string{word}
	}
	matches, err := filepath.Glob(word)
	if err != nil || len(matches) == 0 {
		return []string{word}
	}
	return matches
}

// tokenize splits input into tokens, expands variables, handles quotes and globs.
func (s *Shell) tokenize(input string) []string {
	var rawTokens []string
	current := strings.Builder{}
	inSingle := false
	inDouble := false
	parenDepth := 0   // nesting depth inside $(...) or $((...))
	pendingDollar := false // true after $ when next char is (
	wasQuoted := false

	flush := func() {
		if current.Len() > 0 {
			tok := current.String()
			if wasQuoted {
				tok = "\x00q" + tok
			}
			rawTokens = append(rawTokens, tok)
			current.Reset()
			wasQuoted = false
			pendingDollar = false
		}
	}

	for i := 0; i < len(input); i++ {
		c := input[i]
		switch {
		case c == '\'' && !inDouble && parenDepth == 0:
			inSingle = !inSingle
			wasQuoted = true
			current.WriteByte(c)
		case c == '"' && !inSingle && parenDepth == 0:
			inDouble = !inDouble
			wasQuoted = true
			current.WriteByte(c)
		case c == '$' && !inSingle && i+1 < len(input) && (input[i+1] == '(' || input[i+1] == '{'):
			// Mark that the next ( opens a substitution — don't increment depth here
			if input[i+1] == '(' {
				pendingDollar = true
			}
			current.WriteByte(c)
		case c == '(' && !inSingle && !inDouble && (parenDepth > 0 || pendingDollar):
			parenDepth++
			pendingDollar = false
			current.WriteByte(c)
		case c == ')' && !inSingle && !inDouble && parenDepth > 0:
			parenDepth--
			current.WriteByte(c)
		case c == '{' && !inSingle && !inDouble && parenDepth > 0:
			parenDepth++
			current.WriteByte(c)
		case c == '}' && !inSingle && !inDouble && parenDepth > 0:
			parenDepth--
			current.WriteByte(c)
		case (c == ' ' || c == '\t') && !inSingle && !inDouble && parenDepth == 0:
			flush()
		case c == '#' && !inSingle && !inDouble && parenDepth == 0 && current.Len() == 0:
			// Inline comment: # at start of a new token — discard the rest of the input
			goto doneTokenizing
		default:
			pendingDollar = false
			current.WriteByte(c)
		}
	}
doneTokenizing:
	flush()

	// Handle variable assignment on token[0]: FOO=bar
	if len(rawTokens) > 0 {
		tok := rawTokens[0]
		clean := strings.TrimPrefix(tok, "\x00q")
		if eqIdx := strings.Index(clean, "="); eqIdx > 0 {
			name := clean[:eqIdx]
			if isValidIdentifier(name) && !strings.Contains(clean[:eqIdx], "$") {
				value := s.expandWord(clean[eqIdx+1:])
				s.vars[name] = value
				os.Setenv(name, value)
				rawTokens = rawTokens[1:]
				if len(rawTokens) == 0 {
					return nil
				}
			}
		}
	}

	var result []string
	for _, tok := range rawTokens {
		quoted := strings.HasPrefix(tok, "\x00q")
		if quoted {
			tok = tok[2:]
		}
		expanded := s.expandWord(tok)
		if !quoted && strings.ContainsAny(expanded, "*?[") {
			result = append(result, s.expandGlob(expanded)...)
		} else {
			result = append(result, expanded)
		}
	}
	return result
}