import Foundation

/// Reads text files of unknown provenance.
///
/// `.vcf` and `.ics` files reaching the app may have come from any address book
/// or calendar going back twenty years, so the declared encoding — if there is
/// one — cannot be trusted. Candidates are tried in order of likelihood.
public enum TextFile {
    public static func read(contentsOf url: URL) throws -> String {
        let data = try Data(contentsOf: url)
        return decode(data)
    }

    static func decode(_ data: Data) -> String {
        // A UTF-8 BOM would otherwise survive into the first property name.
        var bytes = data
        if bytes.starts(with: [0xEF, 0xBB, 0xBF]) {
            bytes = bytes.dropFirst(3)
        }

        for encoding in [String.Encoding.utf8, .windowsCP1252, .isoLatin1] {
            if let text = String(data: bytes, encoding: encoding) {
                return text
            }
        }
        return String(decoding: bytes, as: UTF8.self)
    }
}

/// One output file produced by sanitizing.
public struct SanitizedFile: Sendable, Equatable {
    /// Base name without extension; the caller allocates the final filename.
    public var suggestedName: String
    public var contents: String

    public init(suggestedName: String, contents: String) {
        self.suggestedName = suggestedName
        self.contents = contents
    }
}

/// The outcome of sanitizing one input file.
public struct FileSanitizationResult: Sendable, Equatable {
    public var files: [SanitizedFile]
    public var warnings: [String]

    public init(files: [SanitizedFile], warnings: [String] = []) {
        self.files = files
        self.warnings = warnings
    }
}