import Foundation

/// Linear scanning avoids regex backtracking on malformed clipboard HTML such
/// as a long run of '<' without a closing '>'. No rendering or I/O occurs.
struct DocsHTMLScanner {
    enum Piece { case text(String), tag(String) }
    let source: NSString
    var offset = 0
    init(_ value: String) { source = value as NSString }
    mutating func next() -> Piece? {
        guard offset < source.length else { return nil }
        let start = offset
        if source.character(at: offset) != 60 {
            let next = source.range(of: "<", range: NSRange(location: offset, length: source.length - offset))
            offset = next.location == NSNotFound ? source.length : next.location
            return .text(source.substring(with: NSRange(location: start, length: offset - start)))
        }
        if source.length - offset >= 4, source.substring(with: NSRange(location: offset, length: 4)) == "<!--" {
            let end = source.range(of: "-->", range: NSRange(location: offset + 4, length: source.length - offset - 4))
            offset = end.location == NSNotFound ? source.length : NSMaxRange(end)
            return .tag("<!-- -->")
        }
        var quote: unichar = 0
        offset += 1
        while offset < source.length {
            let value = source.character(at: offset)
            offset += 1
            if quote != 0 {
                if value == quote { quote = 0 }
            } else if value == 34 || value == 39 { quote = value }
            else if value == 62 { return .tag(source.substring(with: NSRange(location: start, length: offset - start))) }
        }
        return .text(source.substring(from: start))
    }
    static func attributes(_ value: String) -> [String: String] {
        let source = value as NSString
        var index = 0
        var result: [String: String] = [:]
        let accepted: Set<String> = ["href", "title", "style", "class", "src", "alt", "start", "align", "face"]
        func space(_ value: unichar) -> Bool { [9, 10, 12, 13, 32].contains(value) }
        while index < source.length {
            while index < source.length && space(source.character(at: index)) { index += 1 }
            let start = index
            while index < source.length && !space(source.character(at: index)) && source.character(at: index) != 61 { index += 1 }
            let key = index - start <= 128 ? source.substring(with: NSRange(location: start, length: index - start)).lowercased() : ""
            while index < source.length && space(source.character(at: index)) { index += 1 }
            guard index < source.length else { break }
            guard source.character(at: index) == 61 else { continue }
            index += 1
            while index < source.length && space(source.character(at: index)) { index += 1 }
            guard index < source.length else { break }
            let first = source.character(at: index)
            let quoted = first == 34 || first == 39
            if quoted { index += 1 }
            let content = index
            while index < source.length && (quoted ? source.character(at: index) != first : !space(source.character(at: index))) { index += 1 }
            if accepted.contains(key) { result[key] = source.substring(with: NSRange(location: content, length: index - content)) }
            if quoted && index < source.length { index += 1 }
        }
        return result
    }
}