import Foundation

enum TranscriptCleaner {
  static func clean(_ text: String) -> String {
    var prefix = text.startIndex
    if prefix < text.endIndex, text[prefix] == "\u{FEFF}" {
      prefix = text.index(after: prefix)
    }
    while prefix < text.endIndex, text[prefix].isWhitespace {
      prefix = text.index(after: prefix)
    }
    guard text[prefix...].hasPrefix("language ") else { return text }

    let markers = ["<asr_text>", "<astext>"]
    let marker = markers.compactMap { text.range(of: $0, range: prefix..<text.endIndex) }
      .min { $0.lowerBound < $1.lowerBound }
    guard let marker else { return text }
    return String(text[marker.upperBound...])
  }
}