gosh/expand.go

package main

// Parameter expansion ${...}: ${name}, ${#name} and the operators
//
//	${name:-word} ${name-word}   word if name is unset (or empty with :)
//	${name:=word} ${name=word}   ... and assign it to name
//	${name:+word} ${name+word}   word if name is set (and not empty with :)
//	${name:?word} ${name?word}   error if name is unset (or empty)
//	${name#pat} ${name##pat}     remove the shortest/longest matching prefix
//	${name%pat} ${name%%pat}     remove the shortest/longest matching suffix
//	${name/pat/rep}              replace the first (//: every) match
//	${name/#pat/rep}             ... at the start, /% at the end
//	${name:offset[:length]}      substring (arithmetic expressions)
//
// The word is expanded (variables, $(...), $((...)), quotes, ~), but not
// split. In patterns, quoted characters and values of variables match
// literally, like everywhere in gosh.

import (
	"errors"
	"fmt"
	"os"
	"path"
	"strconv"
	"strings"
	"unicode/utf8"
)

// paramEnd returns the index of the } that ends the ${ at s[i], or -1.
// Nested ${...}, $(...), quotes and escapes are skipped.
func paramEnd(s string, i int) int {
	var quote byte
	for j := i + 2; j < len(s); j++ {
		ch := s[j]
		switch {
		case quote == '\'':
			if ch == '\'' {
				quote = 0
			}
		case ch == '\\':
			j++
		case isSubstStart(s, j):
			end := substEnd(s, j)
			if end < 0 {
				return -1
			}
			j = end
		case ch == '$' && j+1 < len(s) && s[j+1] == '{':
			end := paramEnd(s, j)
			if end < 0 {
				return -1
			}
			j = end
		case quote == '"':
			if ch == '"' {
				quote = 0
			}
		case ch == '"' || ch == '\'':
			quote = ch
		case ch == '}':
			return j
		}
	}
	return -1
}

// paramNameLen returns the length of the parameter name at the start of
// expr: a variable name, a number or a special parameter.
func paramNameLen(expr string) int {
	switch {
	case expr == "":
		return 0
	case isSpecialParam(expr[:1]):
		return 1
	case expr[0] >= '0' && expr[0] <= '9':
		n := 1
		for n < len(expr) && expr[n] >= '0' && expr[n] <= '9' {
			n++
		}
		return n
	}
	n := 0
	for n < len(expr) && isVarNameByte(expr[n], n == 0) {
		n++
	}
	return n
}

// expandParam expands the inside of ${...}.
func expandParam(expr string) (string, error) {
	bad := func() (string, error) { return "", fmt.Errorf("bad substitution: ${%s}", expr) }
	unbound := func(name string) (string, error) { return "", fmt.Errorf("%s: unbound variable", name) }

	// ${#name}: the length in characters (${#} alone is $#)
	if len(expr) > 1 && expr[0] == '#' {
		name := expr[1:]
		if paramNameLen(name) != len(name) {
			return bad()
		}
		value, ok := lookupParam(name)
		if !ok {
			return unbound(name)
		}
		return strconv.Itoa(utf8.RuneCountInString(value)), nil
	}

	n := paramNameLen(expr)
	if n == 0 {
		return bad()
	}
	name, op := expr[:n], expr[n:]
	if op != "" && strings.IndexByte(":-=+?#%/", op[0]) < 0 {
		return bad()
	}
	value, set := lookupParam(name)
	if op == "" {
		if !set {
			return unbound(name)
		}
		return value, nil
	}

	// ${name:-word} and friends
	colon := strings.HasPrefix(op, ":")
	if kind := strings.TrimPrefix(op, ":"); kind != "" && strings.IndexByte("-=+?", kind[0]) >= 0 {
		word := kind[1:]
		missing := !set || colon && value == ""
		switch kind[0] {
		case '-':
			if missing {
				return expandWordText(word)
			}
		case '=':
			if missing {
				if !isVarName(name) {
					return "", fmt.Errorf("$%s: cannot assign in this way", name)
				}
				w, err := expandWordText(word)
				if err != nil {
					return "", err
				}
				os.Setenv(name, w)
				return w, nil
			}
		case '+':
			if missing {
				return "", nil
			}
			return expandWordText(word)
		case '?':
			if missing {
				msg, err := expandWordText(word)
				if err != nil {
					return "", err
				}
				if msg == "" {
					msg = "parameter null or not set"
				}
				return "", fmt.Errorf("%s: %s", name, msg)
			}
		}
		return value, nil
	}

	if !set {
		return unbound(name)
	}
	switch {
	case colon:
		return substring(value, op[1:])
	case strings.HasPrefix(op, "#"), strings.HasPrefix(op, "%"):
		longest := strings.HasPrefix(op[1:], op[:1])
		word := op[1:]
		if longest {
			word = op[2:]
		}
		pattern, err := expandPattern(word)
		if err != nil {
			return "", err
		}
		if op[0] == '#' {
			return removePrefix(value, pattern, longest), nil
		}
		return removeSuffix(value, pattern, longest), nil
	case strings.HasPrefix(op, "/"):
		spec := op[1:]
		all := strings.HasPrefix(spec, "/")
		var anchor byte
		switch {
		case all:
			spec = spec[1:]
		case strings.HasPrefix(spec, "#"), strings.HasPrefix(spec, "%"):
			anchor, spec = spec[0], spec[1:]
		}
		patText, repText := spec, ""
		if i := unquotedSlash(spec); i >= 0 {
			patText, repText = spec[:i], spec[i+1:]
		}
		pattern, err := expandPattern(patText)
		if err != nil {
			return "", err
		}
		rep, err := expandWordText(repText)
		if err != nil {
			return "", err
		}
		return replacePattern(value, pattern, rep, all, anchor), nil
	}
	return bad()
}

// unquotedSlash returns the index of the first / in s that is not quoted,
// escaped or inside $(...) or ${...}, or -1.
func unquotedSlash(s string) int {
	var quote byte
	for i := 0; i < len(s); i++ {
		ch := s[i]
		switch {
		case quote == '\'':
			if ch == '\'' {
				quote = 0
			}
		case ch == '\\':
			i++
		case isSubstStart(s, i):
			if end := substEnd(s, i); end >= 0 {
				i = end
			}
		case ch == '$' && i+1 < len(s) && s[i+1] == '{':
			if end := paramEnd(s, i); end >= 0 {
				i = end
			}
		case quote == '"':
			if ch == '"' {
				quote = 0
			}
		case ch == '"' || ch == '\'':
			quote = ch
		case ch == '/':
			return i
		}
	}
	return -1
}

// substring implements ${name:offset[:length]}; offset and length are
// arithmetic expressions, a negative offset counts from the end.
func substring(value, spec string) (string, error) {
	offText, lenText, hasLen := strings.Cut(spec, ":")
	runes := []rune(value)
	off, err := evalArithText(offText)
	if err != nil {
		return "", err
	}
	if off < 0 {
		off += int64(len(runes))
	}
	if off < 0 || off > int64(len(runes)) {
		return "", nil
	}
	end := int64(len(runes))
	if hasLen {
		n, err := evalArithText(lenText)
		if err != nil {
			return "", err
		}
		if n < 0 {
			end += n // a negative length counts from the end
		} else {
			end = min(off+n, end)
		}
	}
	if end < off {
		return "", errors.New(strings.TrimSpace(lenText) + ": substring expression < 0")
	}
	return string(runes[off:end]), nil
}

// expandWordText expands the word of a parameter expansion (or of $((...)))
// to a single string: quotes are removed, $..., $(...) and $((...)) are
// expanded and a leading ~ is replaced by the home directory.
func expandWordText(word string) (string, error) {
	value, _, err := expandWordParts(word)
	return value, err
}

// expandPattern expands a word like expandWordText, but returns it as a glob
// pattern: quoted characters and expanded values are escaped.
func expandPattern(word string) (string, error) {
	_, pattern, err := expandWordParts(word)
	return pattern, err
}

func expandWordParts(word string) (value, pattern string, err error) {
	var v, p strings.Builder
	literal := func(s string) {
		v.WriteString(s)
		p.WriteString(escapeGlob(s))
	}
	var quote byte
	for i := 0; i < len(word); i++ {
		ch := word[i]
		switch {
		case quote == '\'':
			if ch == '\'' {
				quote = 0
			} else {
				literal(word[i : i+1])
			}
		case ch == '\\' && i+1 < len(word):
			i++
			if quote == '"' && strings.IndexByte("$\"\\`", word[i]) < 0 {
				literal("\\")
			}
			literal(word[i : i+1])
		case ch == '$':
			value, n, err := expandDollar(word[i:])
			if err != nil {
				return "", "", err
			}
			literal(value)
			i += n - 1
		case quote == '"':
			if ch == '"' {
				quote = 0
			} else {
				literal(word[i : i+1])
			}
		case ch == '"' || ch == '\'':
			quote = ch
		case ch == '~' && i == 0 && (len(word) == 1 || word[1] == '/'):
			if home := getHomeDir(); home != "" {
				literal(home)
			} else {
				literal("~")
			}
		default:
			v.WriteByte(ch)
			if ch == '!' && strings.HasSuffix(p.String(), "[") && !strings.HasSuffix(p.String(), "\\[") {
				p.WriteByte('^') // POSIX [!...] is [^...] in Go
			} else {
				p.WriteByte(ch)
			}
		}
	}
	if quote != 0 {
		return "", "", fmt.Errorf("unterminated %c quote", quote)
	}
	return v.String(), p.String(), nil
}

// expandDollar expands the $ expression at the start of s ($name, ${...},
// $(...) or $((...))) and returns its value and length.
func expandDollar(s string) (string, int, error) {
	if end := arithEnd(s, 0); end >= 0 {
		value, err := evalArithText(s[3 : end-1])
		if err != nil {
			return "", 0, err
		}
		return strconv.FormatInt(value, 10), end + 1, nil
	}
	if isSubstStart(s, 0) {
		end := substEnd(s, 0)
		if end < 0 {
			return "", 0, errors.New("missing ) in command substitution")
		}
		out, err := commandSubst(s[2:end])
		return out, end + 1, err
	}
	return expandVar(s)
}

// globMatch reports whether s matches the glob pattern (in which escaped
// characters match literally). Unlike for file names, * and ? also match a
// /. An invalid pattern matches only itself.
func globMatch(pattern, s string) bool {
	// path.Match does not let * match a /; NUL can be in neither string
	toNUL := strings.NewReplacer("/", "\x00")
	ok, err := path.Match(toNUL.Replace(pattern), toNUL.Replace(s))
	if err != nil {
		return unescapeGlob(pattern) == s
	}
	return ok
}

// runeBounds returns the indexes in s where characters start, and len(s).
func runeBounds(s string) []int {
	bounds := make([]int, 0, len(s)+1)
	for i := range s {
		bounds = append(bounds, i)
	}
	return append(bounds, len(s))
}

func removePrefix(value, pattern string, longest bool) string {
	bounds := runeBounds(value)
	for k := range bounds {
		i := bounds[k]
		if longest {
			i = bounds[len(bounds)-1-k]
		}
		if globMatch(pattern, value[:i]) {
			return value[i:]
		}
	}
	return value
}

func removeSuffix(value, pattern string, longest bool) string {
	bounds := runeBounds(value)
	for k := range bounds {
		i := bounds[len(bounds)-1-k]
		if longest {
			i = bounds[k]
		}
		if globMatch(pattern, value[i:]) {
			return value[:i]
		}
	}
	return value
}

// replacePattern replaces the longest match of pattern at the first (or
// every) position by rep. anchor '#' only matches at the start, '%' only at
// the end. Empty matches are not replaced.
func replacePattern(value, pattern, rep string, all bool, anchor byte) string {
	if pattern == "" {
		return value
	}
	bounds := runeBounds(value)
	var b strings.Builder
	for k := 0; k < len(bounds)-1; k++ {
		i := bounds[k]
		match := -1
		if anchor != '#' || i == 0 {
			for m := len(bounds) - 1; m > k; m-- {
				if anchor == '%' && m != len(bounds)-1 {
					break
				}
				if globMatch(pattern, value[i:bounds[m]]) {
					match = m
					break
				}
			}
		}
		if match < 0 {
			b.WriteString(value[i:bounds[k+1]])
			continue
		}
		b.WriteString(rep)
		if !all {
			b.WriteString(value[bounds[match]:])
			return b.String()
		}
		k = match - 1
	}
	return b.String()
}