// Package terminaltext makes untrusted text safe to render in a terminal.
package terminaltext

import "strings"

// Sanitize removes ANSI CSI, OSC (including title/clipboard commands), other
// ESC/C1 control strings, and non-printing controls while preserving printable
// Unicode, newlines, and tabs.
func Sanitize(input string) string {
	const (
		normal = iota
		escape
		csi
		osc
		controlString
		oscEscape
		controlStringEscape
	)

	state := normal
	var output strings.Builder
	output.Grow(len(input))

	for _, current := range input {
		switch state {
		case normal:
			switch current {
			case '\x1b':
				state = escape
			case '\u009b':
				state = csi
			case '\u009d':
				state = osc
			case '\u0090', '\u0098', '\u009e', '\u009f':
				state = controlString
			case '\n', '\t':
				output.WriteRune(current)
			default:
				if current >= ' ' && current != '\x7f' && !(current >= '\u0080' && current <= '\u009f') {
					output.WriteRune(current)
				}
			}
		case escape:
			switch current {
			case '[':
				state = csi
			case ']':
				state = osc
			case 'P', 'X', '^', '_':
				state = controlString
			case '\x1b':
				// Stay in escape state for repeated ESC bytes.
			default:
				state = normal
			}
		case csi:
			if current >= 0x40 && current <= 0x7e {
				state = normal
			}
		case osc:
			switch current {
			case '\a':
				state = normal
			case '\x1b':
				state = oscEscape
			}
		case controlString:
			if current == '\x1b' {
				state = controlStringEscape
			}
		case oscEscape:
			if current == '\\' {
				state = normal
			} else if current != '\x1b' {
				state = osc
			}
		case controlStringEscape:
			if current == '\\' {
				state = normal
			} else if current != '\x1b' {
				state = controlString
			}
		}
	}

	return output.String()
}