wx-win-agent/node-agent/WxAgent.Core/ChatMessages.cs

using System.Security.Cryptography;
using System.Text;

namespace WxAgent.Core;

public enum ChatMessageType
{
    Text,
    Image,
    File,
    Video,
    Voice,
    Link,
    Location,
    Emotion,
    Merge,
    PersonalCard,
    Note,
    Other,
    Quote,
    System
}

public sealed record QuotedMessageSnapshot(string? Sender, string Text);

public sealed record ChatMessageSource(string Text, string? SourceId = null);

public sealed record ChatMessageSnapshot(
    string Text,
    string Fingerprint,
    int VisibleIndex,
    ChatMessageType Type,
    QuotedMessageSnapshot? Quote = null);

public static class VisibleMessageParser
{
    public static IReadOnlyList<ChatMessageSnapshot> Parse(IEnumerable<string> accessibleNames) =>
        Parse(accessibleNames.Select(text => new ChatMessageSource(text)));

    public static IReadOnlyList<ChatMessageSnapshot> Parse(IEnumerable<ChatMessageSource> sources)
    {
        var occurrences = new Dictionary<string, int>(StringComparer.Ordinal);
        var messages = new List<ChatMessageSnapshot>();
        foreach (var source in sources.Where(value => !string.IsNullOrWhiteSpace(value.Text)))
        {
            var text = source.Text;
            var occurrence = occurrences.TryGetValue(text, out var count) ? count : 0;
            occurrences[text] = occurrence + 1;
            var identity = $"{occurrence}:{text}";
            var fingerprint = Convert.ToHexString(SHA256.HashData(Encoding.UTF8.GetBytes(identity))).ToLowerInvariant();
            var (replyText, quote) = ParseQuote(text);
            var type = quote is null ? Classify(replyText) : ChatMessageType.Quote;
            messages.Add(new ChatMessageSnapshot(replyText, fingerprint, messages.Count, type, quote));
        }

        return messages;
    }

    public static (string Text, QuotedMessageSnapshot? Quote) ParseQuote(string accessibleName)
    {
        var normalized = accessibleName.ReplaceLineEndings("\n");
        var lastLine = normalized.LastIndexOf('\n');
        if (lastLine >= 0 && normalized[(lastLine + 1)..].StartsWith("引用 ", StringComparison.Ordinal))
        {
            var descriptor = normalized[(lastLine + 1)..];
            var messageMarker = descriptor.IndexOf(" 的消息", StringComparison.Ordinal);
            var separator = descriptor.IndexOfAny([':', '：']);
            if (messageMarker > 3 && separator > messageMarker)
            {
                var sender = descriptor[3..messageMarker].Trim();
                var quoteText = descriptor[(separator + 1)..].Trim();
                return (normalized[..lastLine], new QuotedMessageSnapshot(sender, quoteText));
            }
        }

        var lines = accessibleName.Split('\n', StringSplitOptions.RemoveEmptyEntries | StringSplitOptions.TrimEntries);
        var quoteIndex = Array.FindIndex(lines, line =>
            string.Equals(line, "引用", StringComparison.Ordinal) || line.StartsWith("引用：", StringComparison.Ordinal));
        if (quoteIndex < 0 || lines.Length - quoteIndex < 3)
        {
            return (accessibleName, null);
        }

        var marker = lines[quoteIndex];
        var legacySender = marker.Length > 3 ? marker[3..].Trim() : null;
        var legacyQuoteText = string.Join('\n', lines[(quoteIndex + 1)..^1]);
        return (lines[^1], new QuotedMessageSnapshot(string.IsNullOrWhiteSpace(legacySender) ? null : legacySender, legacyQuoteText));
    }

    public static ChatMessageType Classify(string text)
    {
        if (text is "[图片]" or "图片" || text.StartsWith("图片\n", StringComparison.Ordinal))
        {
            return ChatMessageType.Image;
        }

        if (text is "[视频]" or "视频" || text.StartsWith("视频\n", StringComparison.Ordinal))
        {
            return ChatMessageType.Video;
        }

        if (text is "[语音]" or "语音" || text.StartsWith("语音\n", StringComparison.Ordinal))
        {
            return ChatMessageType.Voice;
        }

        if (Uri.TryCreate(text, UriKind.Absolute, out var uri) && uri.Scheme is "http" or "https")
        {
            return ChatMessageType.Link;
        }

        if (text.StartsWith("文件\n", StringComparison.Ordinal))
        {
            return ChatMessageType.File;
        }

        if (text is "[位置]" or "位置" || text.StartsWith("位置\n", StringComparison.Ordinal)) return ChatMessageType.Location;
        if (text is "[动画表情]" or "动画表情" or "表情") return ChatMessageType.Emotion;
        if (text.Contains("聊天记录", StringComparison.Ordinal)) return ChatMessageType.Merge;
        if (text is "[名片]" or "个人名片" || text.StartsWith("名片\n", StringComparison.Ordinal)) return ChatMessageType.PersonalCard;
        if (text is "[笔记]" or "笔记" || text.StartsWith("笔记\n", StringComparison.Ordinal)) return ChatMessageType.Note;

        if (!text.Contains(Path.DirectorySeparatorChar) && !text.Contains(Path.AltDirectorySeparatorChar) &&
            Path.GetExtension(text) is { Length: > 1 } extension && extension.Length <= 10)
        {
            return ChatMessageType.File;
        }

        return text.StartsWith("[", StringComparison.Ordinal) && text.EndsWith("]", StringComparison.Ordinal)
            ? ChatMessageType.System
            : ChatMessageType.Text;
    }
}

public sealed class BoundedMessageDeduper
{
    private readonly int _capacity;
    private readonly Queue<string> _order = new();
    private readonly HashSet<string> _seen = new(StringComparer.Ordinal);

    public BoundedMessageDeduper(int capacity = 2048)
    {
        if (capacity <= 0)
            throw new ArgumentOutOfRangeException(nameof(capacity), "Capacity must be positive.");
        _capacity = capacity;
    }

    public bool TryAdd(string fingerprint)
    {
        ArgumentException.ThrowIfNullOrWhiteSpace(fingerprint);
        if (!_seen.Add(fingerprint))
        {
            return false;
        }

        _order.Enqueue(fingerprint);
        if (_order.Count > _capacity)
        {
            _seen.Remove(_order.Dequeue());
        }

        return true;
    }
}