compaction

1 files changed, 46 insertions, 2 deletions

diff --git a/libpanto/src/openai_chat_json.zig b/libpanto/src/openai_chat_json.zig
index 8e6c9be..8035338 100644
--- a/libpanto/src/openai_chat_json.zig
+++ b/libpanto/src/openai_chat_json.zig
@@ -158,8 +158,11 @@ pub fn serializeRequest(
         try s.write(text);
         try s.endObject();
     }
-    // Then every non-system message, in its original order.
-    for (conv.messages.items) |msg| {
+    // Then every non-system message, in its original order. If the
+    // conversation has been compacted, only the latest compaction summary
+    // and the messages after it are active; the superseded prefix is
+    // dropped.
+    for (conversation.activeMessageWindow(conv.messages.items)) |msg| {
         if (msg.role == .system) continue;
         try writeMessage(&s, msg, allocator);
     }
@@ -309,6 +312,9 @@ fn concatTextBlocks(
     for (blocks) |block| {
         switch (block) {
             .Text => |tb| try out.appendSlice(allocator, tb.items),
+            // A compaction summary is the synthetic seed text standing in
+            // for a compacted prefix; emit it as ordinary message text.
+            .CompactionSummary => |cs| try out.appendSlice(allocator, cs.text.items),
             // Thinking, ToolUse, ToolResult: handled elsewhere or dropped.
             else => {},
         }
@@ -945,3 +951,41 @@ test "parseStreamEvent - bare-string error" {
     try testing.expectEqualStrings("something went wrong", pd.delta.error_message.?);
     try testing.expect(pd.delta.error_type == null);
 }
+
+test "serializeRequest - compaction summary trims superseded prefix" {
+    const allocator = testing.allocator;
+
+    var conv = conversation.Conversation.init(allocator);
+    defer conv.deinit();
+
+    // System prompt survives compaction.
+    try conv.addSystemMessage("you are helpful");
+    // Old prefix that must be dropped.
+    try conv.addUserMessage("ancient question");
+    try conv.addAssistantMessage(&.{
+        .{ .Text = try conversation.textualBlockFromSlice(allocator, "ancient answer") },
+    });
+    // Compaction summary resets conversation context.
+    try conv.addCompactionSummary("summary of the ancient exchange");
+    // Kept-verbatim suffix replayed after the summary.
+    try conv.addUserMessage("recent question");
+
+    var tools = emptyTools();
+    defer tools.deinit();
+    const cfg = testConfig("gpt-4o");
+    const body = try serializeRequest(allocator, &cfg, &conv, &tools);
+    defer allocator.free(body);
+
+    var parsed = try std.json.parseFromSlice(std.json.Value, allocator, body, .{});
+    defer parsed.deinit();
+
+    const msgs = parsed.value.object.get("messages").?.array.items;
+    // system + compaction summary (user) + recent question (user) = 3.
+    try testing.expectEqual(@as(usize, 3), msgs.len);
+    try testing.expectEqualStrings("system", msgs[0].object.get("role").?.string);
+    try testing.expectEqualStrings("you are helpful", msgs[0].object.get("content").?.string);
+    try testing.expectEqualStrings("user", msgs[1].object.get("role").?.string);
+    try testing.expectEqualStrings("summary of the ancient exchange", msgs[1].object.get("content").?.string);
+    try testing.expectEqualStrings("user", msgs[2].object.get("role").?.string);
+    try testing.expectEqualStrings("recent question", msgs[2].object.get("content").?.string);
+}