MultiTerm/MultiTerm.Core/Model/MultiFormatString.cs

using MultiTerm.Core.Types;
using System.Collections.ObjectModel;
using System.Text;

namespace MultiTerm.Core.Model;

/// <summary>
/// A string that contains symbols of multiple formats such as a standard character as well as hexadecimal or binary parts.
/// </summary>
public class MultiFormatString: ObservableCollection<IFormattedCharacter>
{
    private const int charsPerHexBundle = 2;
    private const int charsPerBinBundle = 8;

    #region Insertion / Removal of Characters
    /// <summary>
    /// Inserts item at given index after checking if its valid using <see cref="ValidateValue(FormatType, string)"/>.
    /// </summary>
    /// <param name="index">index where to add item, usually items.Count when using Add method.
    /// <param name="item">item to insert</param>
    protected override void InsertItem(int index, IFormattedCharacter item)
    {
        int offset = 0;
        bool insertSpacingAfter = false;
        if(item is FormattedCharacter formattedChar)
        {
            //check if valid, only insert then
            if(ValidateValue(formattedChar.Format, formattedChar.Character) == false)
            {
                return;
            }

            // check if spacing is required and insert
            if(IsSpacingCharacterRequiredBeforeNewCharacter(this, index + offset, formattedChar))
            {
                base.InsertItem(index, new SpacingCharacter());
                offset++;  // insert item with offset of 1
            }

            if(IsSpacingCharacterRequiredAfterNewCharacter(this, index + offset, formattedChar))
            {
                insertSpacingAfter = true;
            }
        }

        base.InsertItem(index + offset, item);

        if (insertSpacingAfter)
        {
            base.InsertItem(index + offset + 1, new SpacingCharacter());
        }
    }

    protected override void RemoveItem(int index)
    {
        // if formatted character => just remove
        base.RemoveItem(index);

        // if previous item is a spacing character => remove it too
        if(this.ElementAt(index-1) is SpacingCharacter)
        {
            base.RemoveAt(index - 1);
        }
    }

    private static bool IsSpacingCharacterRequiredBeforeNewCharacter(IEnumerable<IFormattedCharacter> collection, int indexOfInsertion, IFormattedCharacter newCharacter)
    {
        bool spacingCharRequired = false;

        // no spacing required if new character is a spacing character
        if (newCharacter is SpacingCharacter) { return false; }
        // no handling implemented for items that are not a FormattedCharacter
        if (newCharacter is not FormattedCharacter newFormattedCharacter)
        { throw new NotImplementedException($"'{nameof(IsSpacingCharacterRequiredBeforeNewCharacter)}()' does not implement handling for type {newCharacter.GetType()}"); }

        // switch spacing required by type of format
        spacingCharRequired = newFormattedCharacter.Format switch
        {
            FormatType.Character => CheckSpacingRequiredBecauseChangedFormat(collection, indexOfInsertion, newFormattedCharacter.Format),

            FormatType.Hexadecimal => CheckSpacingRequiredBecauseChangedFormat(collection, indexOfInsertion, newFormattedCharacter.Format) ||
                                      CheckSpacingRequiredBecauseBundleMaxCountReached(collection, indexOfInsertion, charsPerHexBundle, newFormattedCharacter.Format),

            FormatType.Binary => CheckSpacingRequiredBecauseChangedFormat(collection, indexOfInsertion, newFormattedCharacter.Format) ||
                                      CheckSpacingRequiredBecauseBundleMaxCountReached(collection, indexOfInsertion, charsPerBinBundle, newFormattedCharacter.Format),

            _ => throw new NotImplementedException($"'{nameof(IsSpacingCharacterRequiredBeforeNewCharacter)}()' does not implement handling for format {newFormattedCharacter.Format}"),
        };
        return spacingCharRequired;

        static bool CheckSpacingRequiredBecauseChangedFormat(IEnumerable<IFormattedCharacter> collection, int indexOfInsertion, FormatType newCharFormat)
        {
            // check if previous item is of other type
            if (collection.Any())
            {
                int count = 1;
                FormattedCharacter? prevChar = null;
                while (prevChar is null)
                {
                    try
                    {
                        prevChar = collection.ElementAt(indexOfInsertion - count++) as FormattedCharacter;
                        // format of previous is not equal to new => spacer required
                        if (prevChar != null && prevChar.Format != newCharFormat)
                        {
                            return true;
                        }
                    }
                    catch { break; } // end of collection => break while
                }
            }
            return false;
        }

        static bool CheckSpacingRequiredBecauseBundleMaxCountReached(IEnumerable<IFormattedCharacter> collection, int indexOfInsertion, int charsPerBundle, FormatType newCharFormat)
        {
            // spacing required if previous x characters are of same format
            if (collection.Count() >= charsPerBundle)
            {
                try
                {
                    // get last x elements of collection
                    var lastXElements = collection.Take(new Range(new Index(indexOfInsertion - charsPerBundle), new Index(indexOfInsertion)));
                    // get elements of last x elements where the same format is used
                    var sameFormatElements = lastXElements.Where(x => x is FormattedCharacter formChar && formChar.Format == newCharFormat);
                    // elements of same Format are more (impossible) or equal to amount required for bundle
                    if (sameFormatElements.Count() >= charsPerBundle)
                    {
                        return true;
                    }
                }
                catch { } // catch any exception, ignore it and return default
            }
            return false;
        }
    }

    private static bool IsSpacingCharacterRequiredAfterNewCharacter(IEnumerable<IFormattedCharacter> collection, int indexOfInsertion, IFormattedCharacter newCharacter)
    {
        bool spacingCharRequired = false;

        // no spacing required if new character is a spacing character
        if (newCharacter is SpacingCharacter) { return false; }
        // no handling implemented for items that are not a FormattedCharacter
        if (newCharacter is not FormattedCharacter newFormattedCharacter)
        { throw new NotImplementedException($"'{nameof(IsSpacingCharacterRequiredAfterNewCharacter)}()' does not implement handling for type {newCharacter.GetType()}"); }

        // if there is anything in the collection
        if (collection.Any())
        {
            // check if next character in collection is a FormattedCharacter
            try
            {
                var nextChar = collection.ElementAt(indexOfInsertion) as FormattedCharacter;
                // format of next is not equal to new => spacer required
                if (nextChar != null && nextChar.Format != newFormattedCharacter.Format)
                {
                    return true;
                }
            }
            catch { } // catch any exception and return default result
        }

        return spacingCharRequired;
    }
    #endregion

    /// <summary>
    /// Validates wether a certain <paramref name="value"/> is valid for the given <paramref name="format"/>.
    /// </summary>
    /// <returns>true if valid</returns>
    public static bool ValidateValue(FormatType format, string value)
    {
        // invalid if more than one character
        if(value.Length > 1) 
        { return false; }

        // extract character
        var character = value.First();

        switch (format)
        {
            case FormatType.Character:
                // accept only ascii, not extended ascii or unicode
                return char.IsAscii(character);

            case FormatType.Hexadecimal:
                return (character >= '0' && character <= '9' ||
                        character >= 'A' && character <= 'F' ||
                        character >= 'a' && character <= 'f');

            case FormatType.Binary:
                return (character == '0' || character == '1');

            default:
                throw new NotImplementedException($"'{nameof(ValidateValue)}()' does not implement validation for format {format}");
        }
    }

    #region ToString
    /// <summary>
    /// Converts the <see cref="MultiFormatString"/> to a string that only contains Unicode (UTF-16) characters.
    /// </summary>
    /// <returns>converted string</returns>
    public override string ToString()  // TODO Implement Unit Tests!
    {
        StringBuilder stringBuilder = new();
        string hexConversionCharacters = String.Empty, binaryConversionCharacters = String.Empty;

        // Internal function to check and finalize a Hexadecimal Conversion
        // parameter == null will force ending
        void finalizeHexConversion(FormattedCharacter? formattedChar)
        {
            // hex conversion ongoing?
            if (String.IsNullOrEmpty(hexConversionCharacters) == false)
            {
                // ending conversion or reached limit of characters?
                if (formattedChar == null ||
                    formattedChar.Format != FormatType.Hexadecimal ||
                    hexConversionCharacters.Length == charsPerHexBundle)
                {
                    // finalize conversion
                    stringBuilder.Append(FromHexString(hexConversionCharacters));
                    // reset ongoing conversion
                    hexConversionCharacters = String.Empty;
                }
            }
        }

        // Internal function to check and finalize a Binary Conversion
        // parameter == null will force ending
        void finalizeBinaryConversion(FormattedCharacter? formattedChar)
        {
            // binary conversion ongoing?
            if (String.IsNullOrEmpty(binaryConversionCharacters) == false)
            {
                // ending conversion or reached limit of characters?
                if (formattedChar == null ||
                    formattedChar.Format != FormatType.Binary ||
                    binaryConversionCharacters.Length == charsPerBinBundle)
                {
                    // finalize conversion
                    stringBuilder.Append(FromBinaryString(binaryConversionCharacters));
                    // reset ongoing conversion
                    binaryConversionCharacters = String.Empty;
                }
            }
        }

        // go through every character
        foreach (var character in this.Items)
        {
            // can only handle FormattedCharacters
            if (character is not FormattedCharacter formattedCharacter) { continue; }

            // finalize ongoing conversions if there are any
            finalizeHexConversion(formattedCharacter);
            finalizeBinaryConversion(formattedCharacter);

            switch (formattedCharacter.Format)  // switch by format
            {
                case FormatType.Character:
                    stringBuilder.Append(formattedCharacter.Character);
                    break;

                case FormatType.Hexadecimal:
                    hexConversionCharacters += formattedCharacter.Character;
                    break;

                case FormatType.Binary:
                    binaryConversionCharacters += formattedCharacter.Character;
                    break;

                default:
                    throw new NotImplementedException($"'{nameof(ToString)}()' does not implement conversion for format {formattedCharacter.Format}");
            }
        }

        // fully finalize after the loop, finish all unfinished conversions
        finalizeHexConversion(null);
        finalizeBinaryConversion(null);

        return stringBuilder.ToString();
    }

    // from: https://stackoverflow.com/questions/724862/converting-from-hex-to-string
    // test with: returns: "Hello world" for "48656C6C6F20776F726C64"
    // see https://www.fileformat.info/info/charset/UTF-16/list.htm
    private static string FromHexString(string hexString)
    {
        // Added PadLeft so strings with one character do not get ignored
        string internalHexString = hexString.PadLeft(charsPerHexBundle, '0');
        var bytes = new byte[internalHexString.Length / 2];
        for (var i = 0; i < bytes.Length; i++)
        {
            bytes[i] = Convert.ToByte(internalHexString.Substring(i * 2, 2), 16);
        }

        return Encoding.ASCII.GetString(bytes);
    }

    private static string FromBinaryString(string binaryString)
    {
        string internalBinaryString = binaryString.PadLeft(charsPerBinBundle, '0');
        var bytes = new byte[internalBinaryString.Length / 8];
        for (var i = 0; i < bytes.Length; i++)
        {
            bytes[i] = Convert.ToByte(internalBinaryString.Substring(i * 8, 8), 2);
        }

        return Encoding.ASCII.GetString(bytes);
    }

    #endregion
}