♻️ fix: Prevent Instructions from Removal when nearing Max Context (#5516)

* refactor: getMessagesWithinTokenLimit to accept params object

* refactor: always include instructions in payload if provided

* ci: remove obsolete test

* refactor: update logoutUser to accept request object and handle session destruction

* test: enhance getMessagesWithinTokenLimit tests for instruction handling

Author: danny-avila

api/app/clients/AnthropicClient.js

@@ -416,7 +416,7 @@ class AnthropicClient extends BaseClient {
     }
 
     let { context: messagesInWindow, remainingContextTokens } =
-      await this.getMessagesWithinTokenLimit(formattedMessages);
+      await this.getMessagesWithinTokenLimit({ messages: formattedMessages });
 
     const tokenCountMap = orderedMessages
       .slice(orderedMessages.length - messagesInWindow.length)

api/app/clients/BaseClient.js

@@ -347,25 +347,38 @@ class BaseClient {
    * If the token limit would be exceeded by adding a message, that message is not added to the context and remains in the original array.
    * The method uses `push` and `pop` operations for efficient array manipulation, and reverses the context array at the end to maintain the original order of the messages.
    *
-   * @param {Array} _messages - An array of messages, each with a `tokenCount` property. The messages should be ordered from oldest to newest.
-   * @param {number} [maxContextTokens] - The max number of tokens allowed in the context. If not provided, defaults to `this.maxContextTokens`.
-   * @returns {Object} An object with four properties: `context`, `summaryIndex`, `remainingContextTokens`, and `messagesToRefine`.
+   * @param {Object} params
+   * @param {TMessage[]} params.messages - An array of messages, each with a `tokenCount` property. The messages should be ordered from oldest to newest.
+   * @param {number} [params.maxContextTokens] - The max number of tokens allowed in the context. If not provided, defaults to `this.maxContextTokens`.
+   * @param {{ role: 'system', content: text, tokenCount: number }} [params.instructions] - Instructions already added to the context at index 0.
+   * @returns {Promise<{
+   *  context: TMessage[],
+   *  remainingContextTokens: number,
+   *  messagesToRefine: TMessage[],
+   *  summaryIndex: number,
+   * }>} An object with four properties: `context`, `summaryIndex`, `remainingContextTokens`, and `messagesToRefine`.
    *    `context` is an array of messages that fit within the token limit.
    *    `summaryIndex` is the index of the first message in the `messagesToRefine` array.
    *    `remainingContextTokens` is the number of tokens remaining within the limit after adding the messages to the context.
    *    `messagesToRefine` is an array of messages that were not added to the context because they would have exceeded the token limit.
    */
-  async getMessagesWithinTokenLimit(_messages, maxContextTokens) {
+  async getMessagesWithinTokenLimit({ messages: _messages, maxContextTokens, instructions }) {
     // Every reply is primed with <|start|>assistant<|message|>, so we
     // start with 3 tokens for the label after all messages have been counted.
-    let currentTokenCount = 3;
     let summaryIndex = -1;
-    let remainingContextTokens = maxContextTokens ?? this.maxContextTokens;
+    let currentTokenCount = 3;
+    const instructionsTokenCount = instructions?.tokenCount ?? 0;
+    let remainingContextTokens =
+      (maxContextTokens ?? this.maxContextTokens) - instructionsTokenCount;
     const messages = [..._messages];
 
     const context = [];
+
     if (currentTokenCount < remainingContextTokens) {
       while (messages.length > 0 && currentTokenCount < remainingContextTokens) {
+        if (messages.length === 1 && instructions) {
+          break;
+        }
         const poppedMessage = messages.pop();
         const { tokenCount } = poppedMessage;
 
@@ -379,6 +392,11 @@ class BaseClient {
       }
     }
 
+    if (instructions) {
+      context.push(_messages[0]);
+      messages.shift();
+    }
+
     const prunedMemory = messages;
     summaryIndex = prunedMemory.length - 1;
     remainingContextTokens -= currentTokenCount;
@@ -403,27 +421,38 @@ class BaseClient {
     if (instructions) {
       ({ tokenCount, ..._instructions } = instructions);
     }
+
     _instructions && logger.debug('[BaseClient] instructions tokenCount: ' + tokenCount);
-    let payload = this.addInstructions(formattedMessages, _instructions);
-    let orderedWithInstructions = this.addInstructions(orderedMessages, instructions);
+    if (tokenCount && tokenCount > this.maxContextTokens) {
+      const info = `${tokenCount} / ${this.maxContextTokens}`;
+      const errorMessage = `{ "type": "${ErrorTypes.INPUT_LENGTH}", "info": "${info}" }`;
+      logger.warn(`Instructions token count exceeds max token count (${info}).`);
+      throw new Error(errorMessage);
+    }
+
     if (this.clientName === EModelEndpoint.agents) {
       const { dbMessages, editedIndices } = truncateToolCallOutputs(
-        orderedWithInstructions,
+        orderedMessages,
         this.maxContextTokens,
         this.getTokenCountForMessage.bind(this),
       );
 
       if (editedIndices.length > 0) {
         logger.debug('[BaseClient] Truncated tool call outputs:', editedIndices);
         for (const index of editedIndices) {
-          payload[index].content = dbMessages[index].content;
+          formattedMessages[index].content = dbMessages[index].content;
         }
-        orderedWithInstructions = dbMessages;
+        orderedMessages = dbMessages;
       }
     }
 
+    let orderedWithInstructions = this.addInstructions(orderedMessages, instructions);
+
     let { context, remainingContextTokens, messagesToRefine, summaryIndex } =
-      await this.getMessagesWithinTokenLimit(orderedWithInstructions);
+      await this.getMessagesWithinTokenLimit({
+        messages: orderedWithInstructions,
+        instructions,
+      });
 
     logger.debug('[BaseClient] Context Count (1/2)', {
       remainingContextTokens,
@@ -435,7 +464,9 @@ class BaseClient {
     let { shouldSummarize } = this;
 
     // Calculate the difference in length to determine how many messages were discarded if any
-    const { length } = payload;
+    let payload;
+    let { length } = formattedMessages;
+    length += instructions != null ? 1 : 0;
     const diff = length - context.length;
     const firstMessage = orderedWithInstructions[0];
     const usePrevSummary =
@@ -445,18 +476,31 @@ class BaseClient {
       this.previous_summary.messageId === firstMessage.messageId;
 
     if (diff > 0) {
-      payload = payload.slice(diff);
+      payload = formattedMessages.slice(diff);
       logger.debug(
         `[BaseClient] Difference between original payload (${length}) and context (${context.length}): ${diff}`,
       );
     }
 
+    payload = this.addInstructions(payload ?? formattedMessages, _instructions);
+
     const latestMessage = orderedWithInstructions[orderedWithInstructions.length - 1];
     if (payload.length === 0 && !shouldSummarize && latestMessage) {
       const info = `${latestMessage.tokenCount} / ${this.maxContextTokens}`;
       const errorMessage = `{ "type": "${ErrorTypes.INPUT_LENGTH}", "info": "${info}" }`;
       logger.warn(`Prompt token count exceeds max token count (${info}).`);
       throw new Error(errorMessage);
+    } else if (
+      _instructions &&
+      payload.length === 1 &&
+      payload[0].content === _instructions.content
+    ) {
+      const info = `${tokenCount + 3} / ${this.maxContextTokens}`;
+      const errorMessage = `{ "type": "${ErrorTypes.INPUT_LENGTH}", "info": "${info}" }`;
+      logger.warn(
+        `Including instructions, the prompt token count exceeds remaining max token count (${info}).`,
+      );
+      throw new Error(errorMessage);
     }
 
     if (usePrevSummary) {

api/app/clients/OpenAIClient.js

@@ -931,7 +931,10 @@ ${convo}
     );
 
     if (excessTokenCount > maxContextTokens) {
-      ({ context } = await this.getMessagesWithinTokenLimit(context, maxContextTokens));
+      ({ context } = await this.getMessagesWithinTokenLimit({
+        messages: context,
+        maxContextTokens,
+      }));
     }
 
     if (context.length === 0) {

api/app/clients/specs/BaseClient.test.js

@@ -159,7 +159,7 @@ describe('BaseClient', () => {
       expectedMessagesToRefine?.[expectedMessagesToRefine.length - 1] ?? {};
     const expectedIndex = messages.findIndex((msg) => msg.content === lastExpectedMessage?.content);
 
-    const result = await TestClient.getMessagesWithinTokenLimit(messages);
+    const result = await TestClient.getMessagesWithinTokenLimit({ messages });
 
     expect(result.context).toEqual(expectedContext);
     expect(result.summaryIndex).toEqual(expectedIndex);
@@ -195,74 +195,14 @@ describe('BaseClient', () => {
       expectedMessagesToRefine?.[expectedMessagesToRefine.length - 1] ?? {};
     const expectedIndex = messages.findIndex((msg) => msg.content === lastExpectedMessage?.content);
 
-    const result = await TestClient.getMessagesWithinTokenLimit(messages);
+    const result = await TestClient.getMessagesWithinTokenLimit({ messages });
 
     expect(result.context).toEqual(expectedContext);
     expect(result.summaryIndex).toEqual(expectedIndex);
     expect(result.remainingContextTokens).toBe(expectedRemainingContextTokens);
     expect(result.messagesToRefine).toEqual(expectedMessagesToRefine);
   });
 
-  test('handles context strategy correctly in handleContextStrategy()', async () => {
-    TestClient.addInstructions = jest
-      .fn()
-      .mockReturnValue([
-        { content: 'Hello' },
-        { content: 'How can I help you?' },
-        { content: 'Please provide more details.' },
-        { content: 'I can assist you with that.' },
-      ]);
-    TestClient.getMessagesWithinTokenLimit = jest.fn().mockReturnValue({
-      context: [
-        { content: 'How can I help you?' },
-        { content: 'Please provide more details.' },
-        { content: 'I can assist you with that.' },
-      ],
-      remainingContextTokens: 80,
-      messagesToRefine: [{ content: 'Hello' }],
-      summaryIndex: 3,
-    });
-
-    TestClient.getTokenCount = jest.fn().mockReturnValue(40);
-
-    const instructions = { content: 'Please provide more details.' };
-    const orderedMessages = [
-      { content: 'Hello' },
-      { content: 'How can I help you?' },
-      { content: 'Please provide more details.' },
-      { content: 'I can assist you with that.' },
-    ];
-    const formattedMessages = [
-      { content: 'Hello' },
-      { content: 'How can I help you?' },
-      { content: 'Please provide more details.' },
-      { content: 'I can assist you with that.' },
-    ];
-    const expectedResult = {
-      payload: [
-        {
-          role: 'system',
-          content: 'Refined answer',
-        },
-        { content: 'How can I help you?' },
-        { content: 'Please provide more details.' },
-        { content: 'I can assist you with that.' },
-      ],
-      promptTokens: expect.any(Number),
-      tokenCountMap: {},
-      messages: expect.any(Array),
-    };
-
-    TestClient.shouldSummarize = true;
-    const result = await TestClient.handleContextStrategy({
-      instructions,
-      orderedMessages,
-      formattedMessages,
-    });
-
-    expect(result).toEqual(expectedResult);
-  });
-
   describe('getMessagesForConversation', () => {
     it('should return an empty array if the parentMessageId does not exist', () => {
       const result = TestClient.constructor.getMessagesForConversation({
@@ -674,4 +614,112 @@ describe('BaseClient', () => {
       expect(calls[1][0].isCreatedByUser).toBe(false); // Second call should be for response message
     });
   });
+
+  describe('getMessagesWithinTokenLimit with instructions', () => {
+    test('should always include instructions when present', async () => {
+      TestClient.maxContextTokens = 50;
+      const instructions = {
+        role: 'system',
+        content: 'System instructions',
+        tokenCount: 20,
+      };
+
+      const messages = [
+        instructions,
+        { role: 'user', content: 'Hello', tokenCount: 10 },
+        { role: 'assistant', content: 'Hi there', tokenCount: 15 },
+      ];
+
+      const result = await TestClient.getMessagesWithinTokenLimit({
+        messages,
+        instructions,
+      });
+
+      expect(result.context[0]).toBe(instructions);
+      expect(result.remainingContextTokens).toBe(2);
+    });
+
+    test('should handle case when messages exceed limit but instructions must be preserved', async () => {
+      TestClient.maxContextTokens = 30;
+      const instructions = {
+        role: 'system',
+        content: 'System instructions',
+        tokenCount: 20,
+      };
+
+      const messages = [
+        instructions,
+        { role: 'user', content: 'Hello', tokenCount: 10 },
+        { role: 'assistant', content: 'Hi there', tokenCount: 15 },
+      ];
+
+      const result = await TestClient.getMessagesWithinTokenLimit({
+        messages,
+        instructions,
+      });
+
+      // Should only include instructions and the last message that fits
+      expect(result.context).toHaveLength(1);
+      expect(result.context[0].content).toBe(instructions.content);
+      expect(result.messagesToRefine).toHaveLength(2);
+      expect(result.remainingContextTokens).toBe(7); // 30 - 20 - 3 (assistant label)
+    });
+
+    test('should work correctly without instructions (1/2)', async () => {
+      TestClient.maxContextTokens = 50;
+      const messages = [
+        { role: 'user', content: 'Hello', tokenCount: 10 },
+        { role: 'assistant', content: 'Hi there', tokenCount: 15 },
+      ];
+
+      const result = await TestClient.getMessagesWithinTokenLimit({
+        messages,
+      });
+
+      expect(result.context).toHaveLength(2);
+      expect(result.remainingContextTokens).toBe(22); // 50 - 10 - 15 - 3(assistant label)
+      expect(result.messagesToRefine).toHaveLength(0);
+    });
+
+    test('should work correctly without instructions (2/2)', async () => {
+      TestClient.maxContextTokens = 30;
+      const messages = [
+        { role: 'user', content: 'Hello', tokenCount: 10 },
+        { role: 'assistant', content: 'Hi there', tokenCount: 20 },
+      ];
+
+      const result = await TestClient.getMessagesWithinTokenLimit({
+        messages,
+      });
+
+      expect(result.context).toHaveLength(1);
+      expect(result.remainingContextTokens).toBe(7);
+      expect(result.messagesToRefine).toHaveLength(1);
+    });
+
+    test('should handle case when only instructions fit within limit', async () => {
+      TestClient.maxContextTokens = 25;
+      const instructions = {
+        role: 'system',
+        content: 'System instructions',
+        tokenCount: 20,
+      };
+
+      const messages = [
+        instructions,
+        { role: 'user', content: 'Hello', tokenCount: 10 },
+        { role: 'assistant', content: 'Hi there', tokenCount: 15 },
+      ];
+
+      const result = await TestClient.getMessagesWithinTokenLimit({
+        messages,
+        instructions,
+      });
+
+      expect(result.context).toHaveLength(1);
+      expect(result.context[0]).toBe(instructions);
+      expect(result.messagesToRefine).toHaveLength(2);
+      expect(result.remainingContextTokens).toBe(2); // 25 - 20 - 3(assistant label)
+    });
+  });
 });

api/server/controllers/auth/LogoutController.js

@@ -5,7 +5,7 @@ const { logger } = require('~/config');
 const logoutController = async (req, res) => {
   const refreshToken = req.headers.cookie ? cookies.parse(req.headers.cookie).refreshToken : null;
   try {
-    const logout = await logoutUser(req.user._id, refreshToken);
+    const logout = await logoutUser(req, refreshToken);
     const { status, message } = logout;
     res.clearCookie('refreshToken');
     return res.status(status).send({ message });