How to remove session memory via python api

[LeeMoofon0222]

Hi,

I am currently using Langflow and would like to know how to remove session memory via the Python API.

I have searched for similar discussions and tried the following code to delete a session:

def delete_session(session_id):
    api_url = f"{SESSION_URL}/{session_id}"
    headers = {
        "x-api-key": API_KEY
    }
    response = requests.delete(api_url, headers=headers)
    if response.status_code == 200:
        print("Session deleted successfully.")
    else:
        print(f"Failed to delete session. Status code: {response.status_code}, Response: {response.text}")

However, I am receiving the following response:

Failed to delete session. Status code: 405, Response: {"detail":"Method Not Allowed"}

It seems like the DELETE method is not allowed for the session endpoint. Could you please guide me on the correct approach to remove session memory via the API?

Thank you!

[dosubot[bot]]

Hi @LeeMoofon0222! I'm here to help you out. Let's work together to tackle this issue.

To remove session memory via the Langflow Python API, you can use the SessionService class provided in the service.py file. This class includes a method to clear the session for a given session ID. Here is how you can use it:

from langflow.services.session.service import SessionService
from langflow.services.cache.base import CacheService

# Create an instance of the cache service
cache_service = CacheService()

# Create an instance of the session service
session_service = SessionService(cache_service)

# Specify the session ID you want to clear
session_id = "your_session_id_here"

# Call the clear_session method to clear the session memory
await session_service.clear_session(session_id)

# Optionally, you can check if the session has been cleared
cleared_data = cache_service.get(session_id)
print(cleared_data)  # Should print None if the session was successfully cleared

Alternatively, you can use the delete_messages_session endpoint provided by the MonitorService in the Langflow API to delete session messages. Here is an example of how to use it:

import requests

# Define the session ID you want to clear
session_id = "your_session_id_here"

# Define the URL for the delete_messages_session endpoint
url = f"http://your-langflow-api-url/monitor/messages/session/{session_id}"

# Make the DELETE request to clear the session messages
response = requests.delete(url)

# Check the response status
if response.status_code == 204:
    print("Session messages successfully cleared.")
else:
    print(f"Failed to clear session messages: {response.status_code} - {response.text}")

This code will clear the session memory for the specified session ID using the delete_messages_session endpoint within the MonitorService class ^[1][2][3].

To continue talking to Dosu, mention @dosu.

[dosubot[bot]]

To call the Langflow API via a URL to load session memory using the MonitorService, you can use the /monitor/messages endpoint. Here is an example of how you can construct the URL:

http://<your-server-address>/monitor/messages?session_id=<your-session-id>&sender=<sender>&sender_name=<sender-name>&order_by=timestamp

Replace <your-server-address>, <your-session-id>, <sender>, and <sender-name> with the appropriate values.

For example:

http://localhost:8000/monitor/messages?session_id=12345&sender=user&sender_name=JohnDoe&order_by=timestamp

This URL will call the get_messages function in the MonitorService to load session memory based on the provided query parameters ^[1].

To continue talking to Dosu, mention @dosu.

[LeeMoofon0222]

@dosu

def load_session(session_id):
    url = f"http://127.0.0.1:7860/monitor/messages?session_id={session_id}&sender=user&sender_name=User&order_by=timestamp"

    response = requests.get(url)
    print(response)
    

print out:
<Response [200]>

I have no idea how to print the messenges out from the session.

[dosubot[bot]]

To print the messages from the session, you need to extract the JSON content from the response and then iterate over the messages. Here is how you can do it:

import requests

def load_session(session_id):
    url = f"http://127.0.0.1:7860/monitor/messages?session_id={session_id}&sender=user&sender_name=User&order_by=timestamp"
    
    response = requests.get(url)
    
    if response.status_code == 200:
        messages = response.json()
        for message in messages:
            print(f"Sender: {message['sender']}, Text: {message['message']}, Session ID: {message['session_id']}")
    else:
        print(f"Failed to retrieve messages: {response.status_code} - {response.text}")

# Example usage
load_session("your_session_id_here")

Replace "your_session_id_here" with the actual session ID you want to load messages for. This code will print out each message's sender, text, and session ID ^{[[1]](https://github.com/langflow-ai/langflow/blob/dev/src/backend/base/langflow/initial_setup/starter_projects/Langflow Memory Conversation.json)[2]}.

To continue talking to Dosu, mention @dosu.

[LeeMoofon0222]

@dosu
messages = response.json()
JSONDecodeError: Expecting value: line 1 column 1 (char 0)

The response text is just a short string <Response [200]>

[dosubot[bot]]

The JSONDecodeError: Expecting value: line 1 column 1 (char 0) error typically occurs when the response content is empty or not a valid JSON. To handle this, you should first check if the response content is empty before attempting to parse it as JSON.

Here's an updated version of your code that includes this check:

import requests

def load_session(session_id):
    url = f"http://127.0.0.1:7860/monitor/messages?session_id={session_id}&sender=user&sender_name=User&order_by=timestamp"
    
    response = requests.get(url)
    
    if response.status_code == 200:
        if response.text:
            try:
                messages = response.json()
                for message in messages:
                    print(f"Sender: {message['sender']}, Text: {message['message']}, Session ID: {message['session_id']}")
            except json.JSONDecodeError as e:
                print(f"JSONDecodeError: {e}")
        else:
            print("Response content is empty.")
    else:
        print(f"Failed to retrieve messages: {response.status_code} - {response.text}")

# Example usage
load_session("your_session_id_here")

This code checks if the response content is not empty before attempting to parse it as JSON, which should help avoid the JSONDecodeError ^{[[1]](https://github.com/langflow-ai/langflow/blob/dev/src/backend/base/langflow/initial_setup/starter_projects/Langflow Memory Conversation.json)[2]}.

To continue talking to Dosu, mention @dosu.

[LeeMoofon0222]

def delete_session(session_id):

    url = f"http://127.0.0.1:7860/api/v1/monitor/messages/session/{session_id}"

    response = requests.delete(url)

    if response.status_code != 204:
        print(f"Failed to clear session messages: {response.status_code} - {response.text}")

It works! Thank you.