wordeck-native/Sources/Core/Domain/CSVParser.swift

import Foundation

/// CSV-Zeile aus dem Import-Flow. `type` ist optional — fehlt es,
/// wird `.basic` angenommen.
struct CSVRow: Equatable {
    let front: String
    let back: String
    let type: CardType

    init(front: String, back: String, type: CardType = .basic) {
        self.front = front
        self.back = back
        self.type = type
    }
}

/// Pragmatischer CSV-Parser für den Cards-Import. Format pro Zeile:
///
///     <question>,<answer>[,<type>]
///
/// - Quote-Escape via `""` (RFC-4180).
/// - Felder dürfen Kommas und Newlines enthalten, wenn sie in `"…"`
///   gekapselt sind.
/// - Header-Row wird automatisch übersprungen, wenn Front/Back beide
///   wie Header-Tokens aussehen (`front`, `back`, `question`, `answer`,
///   `vorderseite`, `rückseite` …).
/// - BOM (`\u{FEFF}`) am Anfang wird gestrippt.
/// - `type` darf jede Wordeck-Type-Bezeichnung sein; unbekannte Werte
///   landen als `.basic`.
enum CSVParser {
    enum ParseError: LocalizedError {
        case empty
        case noValidRows

        var errorDescription: String? {
            switch self {
            case .empty: "Datei ist leer."
            case .noValidRows: "Keine gültigen Zeilen gefunden — erwartet ‚vorne,hinten[,typ]'."
            }
        }
    }

    static func parse(_ rawText: String) throws -> [CSVRow] {
        var text = rawText
        if text.hasPrefix("\u{FEFF}") {
            text.removeFirst()
        }
        if text.trimmingCharacters(in: .whitespacesAndNewlines).isEmpty {
            throw ParseError.empty
        }

        let allRows = parseFields(text)
        guard !allRows.isEmpty else { throw ParseError.noValidRows }

        // Header-Detection: erste Zeile droppen wenn beide Felder Headerwords sind.
        let headerTokens: Set = [
            "front", "back", "question", "answer",
            "vorderseite", "rückseite", "rueckseite", "frage", "antwort"
        ]
        var rows = allRows
        if let first = rows.first,
           first.count >= 2,
           headerTokens.contains(first[0].lowercased()),
           headerTokens.contains(first[1].lowercased())
        {
            rows.removeFirst()
        }

        let parsed: [CSVRow] = rows.compactMap { fields in
            guard fields.count >= 2 else { return nil }
            let front = fields[0].trimmingCharacters(in: .whitespacesAndNewlines)
            let back = fields[1].trimmingCharacters(in: .whitespacesAndNewlines)
            if front.isEmpty, back.isEmpty { return nil }
            let type: CardType = fields.count >= 3
                ? CardType(rawValue: fields[2].trimmingCharacters(in: .whitespacesAndNewlines)) ?? .basic
                : .basic
            return CSVRow(front: front, back: back, type: type)
        }

        if parsed.isEmpty {
            throw ParseError.noValidRows
        }
        return parsed
    }

    /// Parser-State-Machine: liest Zeichen-für-Zeichen, beachtet Quote-
    /// Modus für Kommas/Newlines innerhalb von `"…"`-Feldern. `""` wird
    /// als wörtliches `"` im Feld behandelt.
    private static func parseFields(_ text: String) -> [[String]] {
        var state = ParseState()
        var iterator = text.makeIterator()

        while let char = iterator.next() {
            if state.inQuotes {
                handleQuotedChar(char, iterator: &iterator, state: &state)
            } else if char == "\"", state.currentField.isEmpty {
                state.inQuotes = true
            } else {
                handleUnquotedChar(char, state: &state)
            }
        }

        // Tail-Flush — letzte Zeile ohne abschließendes Newline.
        if !state.currentField.isEmpty || !state.currentRow.isEmpty {
            state.currentRow.append(state.currentField)
            state.rows.append(state.currentRow)
        }

        return state.rows
    }

    /// Mutable State der Parse-Machine — als `inout`-Struct in die
    /// Char-Handler durchgereicht, damit die Parameter-Listen kompakt
    /// bleiben.
    fileprivate struct ParseState {
        var rows: [[String]] = []
        var currentRow: [String] = []
        var currentField = ""
        var inQuotes = false
    }

    /// Im Quote-Modus: `"` schließt das Feld oder escaped sich selbst,
    /// alles andere ist Inhalt.
    private static func handleQuotedChar(
        _ char: Character,
        iterator: inout String.Iterator,
        state: inout ParseState
    ) {
        guard char == "\"" else {
            state.currentField.append(char)
            return
        }
        if let next = iterator.next(), next == "\"" {
            state.currentField.append("\"")
            return
        }
        state.inQuotes = false
        // Das Zeichen nach dem End-Quote ist ein Separator (Komma/Newline
        // /EOF) — über den Unquoted-Handler routen.
        if let next = iterator.next() {
            handleUnquotedChar(next, state: &state)
        }
    }

    private static func handleUnquotedChar(_ char: Character, state: inout ParseState) {
        switch char {
        case ",":
            state.currentRow.append(state.currentField)
            state.currentField = ""
        case "\n":
            state.currentRow.append(state.currentField)
            state.rows.append(state.currentRow)
            state.currentField = ""
            state.currentRow = []
        case "\r":
            // CRLF: `\r` schluken, `\n` macht den Row-Break.
            break
        case "\"" where state.currentField.isEmpty:
            state.inQuotes = true
        default:
            state.currentField.append(char)
        }
    }
}