AnchorCli/ContextCompactor.cs

using Microsoft.Extensions.AI;
using System.Text.RegularExpressions;
using Spectre.Console;

namespace AnchorCli;

/// <summary>
/// Compacts the chat history when context usage grows too large.
/// Asks the same model to summarize the conversation, then replaces
/// the history with [system prompt, summary, last N user/assistant turns].
/// </summary>
internal sealed partial class ContextCompactor(IChatClient client)
{
    /// <summary>Number of recent user+assistant turn pairs to keep verbatim.</summary>
    private const int KeepRecentTurns = 2;

    /// <summary>Minimum result length to consider for compaction.</summary>
    private const int MinResultLength = 300;

    /// <summary>Matches hashline-encoded output: "lineNumber:hash|content"</summary>
    private static readonly Regex HashlinePattern =
        MyRegex();

    private readonly IChatClient _client = client;

    /// <summary>
    /// Compacts large hashline-encoded tool results from previous turns.
    /// This is the cheapest and most impactful optimization — no LLM call needed.
    /// A 300-line ReadFile result (~10K tokens) becomes a one-line note (~20 tokens).
    /// </summary>
    /// <param name="history">The chat history to compact in-place.</param>
    /// <returns>Number of tool results that were compacted.</returns>
    public static int CompactStaleToolResults(List<ChatMessage> history)
    {
        int compacted = 0;
        int userTurnsSeen = 0;
        Dictionary<string, int> filesRead = new(StringComparer.OrdinalIgnoreCase);

        // Walk backwards: index 0 is system prompt, so we stop at 1.
        for (int i = history.Count - 1; i >= 1; i--)
        {
            var msg = history[i];

            if (msg.Role == ChatRole.User)
            {
                userTurnsSeen++;
            }
            else if (msg.Role == ChatRole.Assistant)
            {
                // Find all FunctionCalls in this assistant message to map call IDs to file paths
                var calls = msg.Contents.OfType<FunctionCallContent>().ToList();

                // If the very next message is a System/Tool role containing FunctionResults
                if (i + 1 < history.Count && history[i + 1].Role == ChatRole.Tool)
                {
                    var resultMsg = history[i + 1];
                    foreach (var content in resultMsg.Contents)
                    {
                        if (content is FunctionResultContent frc &&
                            frc.Result is string resultStr &&
                            resultStr.Length >= MinResultLength &&
                            HashlinePattern.IsMatch(resultStr))
                        {
                            // Find the corresponding function call to check its name/arguments
                            var call = calls.FirstOrDefault(c => c.CallId == frc.CallId);
                            if (call?.Name == "ReadFile" && call.Arguments != null && call.Arguments.TryGetValue("path", out var pathObj))
                            {
                                string filePath = pathObj?.ToString() ?? "";

                                bool shouldRedact = false;
                                string reason = "";

                                // Rule 1: Deduplication. If we have already seen this file in a newer message (since we are walking backward), redact this one.
                                if (filesRead.TryGetValue(filePath, out int count) && count >= 3)
                                {
                                    shouldRedact = true;
                                    reason = "deduplication — you read this file 5 or more times later";
                                }
                                // Rule 2: TTL. If this was read 2 or more user turns ago, redact it.
                                else if (userTurnsSeen >= 2)
                                {
                                    shouldRedact = true;
                                    reason = "expired — read over 2 turns ago";
                                }

                                if (shouldRedact)
                                {
                                    int lineCount = resultStr.Count(c => c == '\n');
                                    frc.Result = $"[File content: {lineCount} lines redacted for {reason}. Re-read the file if you need fresh anchors.]";
                                    compacted++;
                                }
                                else
                                {
                                    // Keep it, but mark that we've seen it so older reads of the same file are redacted.
                                    filesRead[filePath] = filesRead.GetValueOrDefault(filePath) + 1;
                                }
                            }
                        }
                    }
                }
            }
        }

        return compacted;
    }

    /// <summary>
    /// Compacts the history in-place via LLM summarization. Returns true if compaction was performed.
    /// </summary>
    public async Task<bool> TryCompactAsync(
        List<ChatMessage> history,
        CancellationToken ct = default)
    {
        // Need at least: system + some conversation to compact
        if (history.Count < 5)
            return false;

        // Split: system prompt (index 0) | middle (compactable) | tail (keep verbatim)
        var systemMsg = history[0];

        // Find the cut point: keep the last N user+assistant pairs
        int keepFromIndex = FindKeepIndex(history);
        if (keepFromIndex <= 1)
            return false; // Nothing to compact

        // Extract the middle section to summarize
        var toSummarize = history.Skip(1).Take(keepFromIndex - 1).ToList();
        var tail = history.Skip(keepFromIndex).ToList();

        // Build a summarization prompt
        var summaryMessages = new List<ChatMessage>
        {
            new(ChatRole.System, """
                You are a conversation summarizer. Summarize the following coding conversation
                between a user and an AI assistant. Focus on:
                - What files were read, created, or modified (and their paths)
                - What changes were made and why
                - Any decisions or preferences expressed by the user
                - Current state of the work (what's done, what's pending)
                Be concise but preserve all actionable context. Output a single summary paragraph.
                Do NOT include any hashline anchors (lineNumber:hash|) in your summary.
            """)
        };
        summaryMessages.AddRange(toSummarize);
        summaryMessages.Add(new(ChatRole.User,
            "Summarize the above conversation concisely, preserving all important context for continuing the work."));

        string summary;
        try
        {
            var response = await _client.GetResponseAsync(summaryMessages, cancellationToken: ct);
            summary = response.Text ?? "(summary unavailable)";
        }
        catch
        {
            // If summarization fails, don't compact
            return false;
        }

        // Rebuild history: system + summary + recent turns
        history.Clear();
        history.Add(systemMsg);
        history.Add(new(ChatRole.User,
            "[CONTEXT COMPACTED — The following is a summary of the earlier conversation]\n" + summary));
        history.Add(new(ChatRole.Assistant,
            "Understood, I have the context from our earlier conversation. I'll continue from where we left off."));
        history.AddRange(tail);

        return true;
    }

    /// <summary>
    /// Finds the index from which to keep messages verbatim (the last N turn pairs).
    /// </summary>
    private static int FindKeepIndex(List<ChatMessage> history)
    {
        int pairsFound = 0;
        int idx = history.Count - 1;

        while (idx > 0 && pairsFound < KeepRecentTurns)
        {
            // Walk backwards looking for user+assistant pairs
            if (history[idx].Role == ChatRole.Assistant && idx > 1 &&
                history[idx - 1].Role == ChatRole.User)
            {
                pairsFound++;
                idx -= 2;
            }
            else
            {
                idx--;
            }
        }

        // idx+1 is the first message to keep
        return Math.Max(1, idx + 1);
    }

    [GeneratedRegex(@"^\d+:[0-9a-fA-F]{2}\|", RegexOptions.Multiline | RegexOptions.Compiled)]
    private static partial Regex MyRegex();
}