convy/internal/manifest/editlist.go

package manifest

import (
	"fmt"
	"regexp"
	"slices"
	"strings"
)

// A subscription is an array rather than a key-value pair, and the project
// manifest is edited as text for the same reason the suite one is: what the
// author wrote around the data — which component is what, why a topic is taken
// — has to survive the tool touching the file.
//
// So the array is read as the manifest wrote it, not as a regular expression
// over the raw lines: a # inside a string does not open a comment, a bracket
// inside a comment does not close an array, and a name inside a comment is not
// a subscription. Getting any of the three wrong turns a comment into data, and
// there is no way back from that.
//
// The shape survives too. An array written on one line stays on one line, one
// written down a column stays a column, and the comments keep their places: a
// comment block above a value belongs to that value — the same rule that
// governs a comment above a table — so sorting carries it along, while a
// comment trailing on the same line stays on its line.

var arrayKeyRe = regexp.MustCompile(`^(\s*)("[^"]+"|[A-Za-z0-9_-]+)\s*=\s*\[`)

// element is one value of an array together with what was written around it.
type element struct {
	value string
	above []string
	after string
}

// array is an array of the manifest, parsed.
type array struct {
	indent string
	key    string
	// column says the array was written down a column rather than on one line.
	column bool
	elems  []element
	// opening is a comment trailing the opening bracket.
	opening string
	// dangling holds comment lines standing after the last value.
	dangling []string
	// tail is whatever follows the closing bracket.
	tail string
}

// AddToList adds a value to the array under key in the given table. A table
// that has no such key gets one holding the single value.
func AddToList(source []byte, table, key, value string) ([]byte, error) {
	lines := strings.Split(string(source), "\n")

	start, end, ok := tableBounds(lines, table)
	if !ok {
		return nil, fmt.Errorf("the manifest holds no table [%s]", table)
	}

	from, found := arrayKeyLine(lines, start+1, end, key)
	if !found {
		keys, at := tableKeys(lines, start+1, end)
		insert := insertionPoint(keys, at, key, lines, start, end)
		entry := fmt.Sprintf(`%s = ["%s"]`, key, escape(value))
		out := make([]string, 0, len(lines)+1)
		out = append(out, lines[:insert]...)
		out = append(out, entry)
		out = append(out, lines[insert:]...)
		return []byte(strings.Join(out, "\n")), nil
	}

	a, last, ok := readArray(lines, from, end)
	if !ok {
		return nil, fmt.Errorf("the array %s in [%s] is not closed by a bracket", key, table)
	}
	if slices.ContainsFunc(a.elems, func(e element) bool { return e.value == value }) {
		return nil, fmt.Errorf("%s already holds %q", key, value)
	}

	sorted := slices.IsSortedFunc(a.elems, byValue)
	a.elems = append(a.elems, element{value: value})
	if sorted {
		slices.SortStableFunc(a.elems, byValue)
	}
	return splice(lines, from, last, a.render()), nil
}

// RemoveFromList drops a value from the array under key. It is the other half
// of AddToList: unsubscribing is not a command yet, and a pair of edits where
// only one direction is written is a pair where the untried direction is wrong.
func RemoveFromList(source []byte, table, key, value string) ([]byte, error) {
	lines := strings.Split(string(source), "\n")

	start, end, ok := tableBounds(lines, table)
	if !ok {
		return nil, fmt.Errorf("the manifest holds no table [%s]", table)
	}
	from, found := arrayKeyLine(lines, start+1, end, key)
	if !found {
		return nil, fmt.Errorf("the table [%s] holds no key %s", table, key)
	}
	a, last, ok := readArray(lines, from, end)
	if !ok {
		return nil, fmt.Errorf("the array %s in [%s] is not closed by a bracket", key, table)
	}

	i := slices.IndexFunc(a.elems, func(e element) bool { return e.value == value })
	if i < 0 {
		return nil, fmt.Errorf("%s does not hold %q", key, value)
	}
	// The comment above a value went with it and goes away with it; a comment
	// left hanging over the next value would say the wrong thing about it.
	a.elems = slices.Delete(a.elems, i, i+1)
	return splice(lines, from, last, a.render()), nil
}

func byValue(a, b element) int { return strings.Compare(a.value, b.value) }

// AddTable appends a table to the end of the manifest. A new component is a new
// table, and it goes last because the order of components is the author's:
// there is nothing to sort them by that would mean anything.
func AddTable(source []byte, table string, entries [][2]string) ([]byte, error) {
	lines := strings.Split(string(source), "\n")
	if _, _, ok := tableBounds(lines, table); ok {
		return nil, fmt.Errorf("the manifest already holds the table [%s]", table)
	}

	for len(lines) > 0 && strings.TrimSpace(lines[len(lines)-1]) == "" {
		lines = lines[:len(lines)-1]
	}
	block := []string{"", "[" + table + "]"}
	for _, e := range entries {
		block = append(block, e[0]+" = "+e[1])
	}
	block = append(block, "")
	return []byte(strings.Join(append(lines, block...), "\n")), nil
}

// tableBounds finds the lines a table spans: its header and the line the next
// table starts on.
func tableBounds(lines []string, table string) (start, end int, ok bool) {
	start = -1
	for i, line := range lines {
		if m := tableRe.FindStringSubmatch(line); m != nil && m[1] == table {
			start = i
			break
		}
	}
	if start < 0 {
		return 0, 0, false
	}
	end = len(lines)
	for i := start + 1; i < len(lines); i++ {
		if tableRe.MatchString(lines[i]) {
			end = i
			break
		}
	}
	return start, end, true
}

// arrayKeyLine finds the line an array opens on.
func arrayKeyLine(lines []string, from, to int, key string) (int, bool) {
	for i := from; i < to; i++ {
		code, _ := splitComment(lines[i])
		m := arrayKeyRe.FindStringSubmatch(code)
		if m != nil && strings.Trim(m[2], `"`) == key {
			return i, true
		}
	}
	return 0, false
}

// readArray parses the array opening on line from and returns the line it
// closes on.
func readArray(lines []string, from, to int) (a array, last int, ok bool) {
	code, comment := splitComment(lines[from])
	m := arrayKeyRe.FindStringSubmatch(code)
	if m == nil {
		return a, 0, false
	}
	a.indent, a.key = m[1], strings.Trim(m[2], `"`)

	open := strings.Index(code, "[")
	values, depth, closeAt := scanCode(code[open:], 0)
	for _, v := range values {
		a.elems = append(a.elems, element{value: v})
	}
	if closeAt >= 0 {
		a.tail = joinTail(code[open+closeAt:], comment)
		return a, from, true
	}
	a.column = true
	a.opening = comment

	var pending []string
	for n := from + 1; n < to; n++ {
		code, comment := splitComment(lines[n])
		values, next, closeAt := scanCode(code, depth)
		depth = next

		if len(values) == 0 && strings.TrimSpace(code) == "" && closeAt < 0 {
			if comment != "" {
				pending = append(pending, lines[n])
			}
			continue
		}
		for i, v := range values {
			e := element{value: v}
			if i == 0 {
				e.above, pending = pending, nil
			}
			if i == len(values)-1 && closeAt < 0 {
				e.after = comment
			}
			a.elems = append(a.elems, e)
		}
		if closeAt >= 0 {
			a.dangling = pending
			a.tail = joinTail(code[closeAt:], comment)
			return a, n, true
		}
	}
	return a, 0, false
}

// render writes the array back in the shape it had.
func (a array) render() []string {
	quoted := make([]string, len(a.elems))
	for i, e := range a.elems {
		quoted[i] = `"` + escape(e.value) + `"`
	}

	if !a.column {
		line := fmt.Sprintf("%s%s = [%s]", a.indent, a.key, strings.Join(quoted, ", "))
		return []string{appendTail(line, a.tail)}
	}

	out := []string{appendTail(a.indent+a.key+" = [", a.opening)}
	for i, e := range a.elems {
		out = append(out, e.above...)
		out = append(out, appendTail(a.indent+"  "+quoted[i]+",", e.after))
	}
	out = append(out, a.dangling...)
	return append(out, appendTail(a.indent+"]", a.tail))
}

func appendTail(line, tail string) string {
	if tail == "" {
		return line
	}
	return line + " " + tail
}

func joinTail(rest, comment string) string {
	return strings.TrimSpace(strings.TrimSpace(rest) + " " + comment)
}

// splitComment cuts a line into its code and its comment. A # inside a string
// opens no comment, which is the whole reason this is not a call to strings.Cut.
func splitComment(line string) (code, comment string) {
	quote := byte(0)
	for i := 0; i < len(line); i++ {
		c := line[i]
		if quote != 0 {
			if quote == '"' && c == '\\' {
				i++
				continue
			}
			if c == quote {
				quote = 0
			}
			continue
		}
		switch c {
		case '"', '\'':
			quote = c
		case '#':
			return line[:i], line[i:]
		}
	}
	return line, ""
}

// scanCode walks the code of a line, collecting the strings in it and following
// the bracket depth. closeAt is the offset just past the bracket that brought
// the depth back to zero, or -1 while the array is still open.
func scanCode(code string, depth int) (values []string, depthOut, closeAt int) {
	closeAt = -1
	for i := 0; i < len(code); i++ {
		switch c := code[i]; c {
		case '"', '\'':
			var b strings.Builder
			j := i + 1
			for j < len(code) {
				if c == '"' && code[j] == '\\' && j+1 < len(code) {
					b.WriteString(code[j : j+2])
					j += 2
					continue
				}
				if code[j] == c {
					break
				}
				b.WriteByte(code[j])
				j++
			}
			text := b.String()
			if c == '"' {
				text = unescape(text)
			}
			values = append(values, text)
			i = j
		case '[':
			depth++
		case ']':
			depth--
			if depth == 0 && closeAt < 0 {
				closeAt = i + 1
			}
		}
	}
	return values, depth, closeAt
}

// splice replaces lines from..to inclusive with the given block.
func splice(lines []string, from, to int, block []string) []byte {
	out := make([]string, 0, len(lines)+len(block))
	out = append(out, lines[:from]...)
	out = append(out, block...)
	if to < len(lines) {
		out = append(out, lines[to+1:]...)
	}
	return []byte(strings.Join(out, "\n"))
}

func unescape(s string) string {
	s = strings.ReplaceAll(s, `\"`, `"`)
	return strings.ReplaceAll(s, `\\`, `\`)
}