add open func recoverFromLengthy

Sources/LLM/LLM.swift

@@ -205,23 +205,28 @@ open class LLM: ObservableObject {
     private var currentCount: Int32!
     private var decoded = ""
 
+    open func recoverFromLengthy(_ input: borrowing String, to output:  borrowing AsyncStream<String>.Continuation) {
+        output.yield("tl;dr")
+    }
+
     private func prepare(from input: borrowing String, to output: borrowing AsyncStream<String>.Continuation) -> Bool {
         guard !input.isEmpty else { return false }
         context = .init(model, params)
         var tokens = encode(input)
         var initialCount = tokens.count
         currentCount = Int32(initialCount)
         if maxTokenCount <= currentCount {
-            if history.isEmpty {
-                isFull = true
-                output.yield("Input is too long.")
-                return false
-            } else {
+            while !history.isEmpty {
                 history.removeFirst(min(2, history.count))
                 tokens = encode(preProcess(self.input, history))
                 initialCount = tokens.count
                 currentCount = Int32(initialCount)
             }
+            if maxTokenCount <= currentCount {
+                isFull = true
+                recoverFromLengthy(input, to: output)
+                return false
+            }
         }
         for (i, token) in tokens.enumerated() {
             batch.n_tokens = Int32(i)

Tests/LLMTests/LLMTests.swift

@@ -212,4 +212,11 @@ final class LLMTests: XCTestCase {
         await bot.respond(to: input)
         #assert(!bot.output.isEmpty)
     }
+
+    func testRecoveryFromLengtyInput() async throws {
+        var bot = try await LLM(from: model, maxTokenCount: 16)
+        let input = "have you heard of this so-called LLM.swift library?"
+        await bot.respond(to: input)
+        #assert(bot.output == "tl;dr")
+    }
 }