captivate-ai-api 0.3.4

An API for Captivate conversation and state management

Captivate AI & LLM API

Overview

This API is developed by CaptivateChat to handle its API formats.This flexible messaging and metadata management system built using Pydantic models, designed to handle complex communication scenarios with robust type checking and validation.

Key Components

Models

• Captivate: Primary model managing conversation state
• CaptivateResponseModel: Handles response messages and metadata
• ActionModel: Manages actions with flexible payload handling
• ChannelMetadataModel: Stores dynamic channel and conversation metadata

Features

• Dynamic metadata handling
• Immutable session and chat properties
• Flexible message type support
• Custom metadata manipulation
• Conversation title management

You can install through:

pip install captivate-ai-api

Captivate Payload

Here's the JSON payload you will send in the POST request:

{
    "session_id": "lance_catcher_test_69c35e3e-7ff4-484e-8e36-792a62567b79",
    "endpoint": "action",
    "user_input": "hi man",
    "incoming_action": [
        {
            "id": "sendEmail",
            "payload": {
                "email": "[e-mail]",
                "message": "You are fired"
            }
        }
    ],
    "metadata": {
        "internal": {
            "channelMetadata": {
                "course_id": "abc",
                "channelMetadata": {
                    "channel": "custom-channel",
                    "channelData": {}
                },
                "user": {
                    "firstName": "Lance",
                    "lastName": "safa",
                    "email": "[e-mail]"
                },
                "phoneNumber": null,
                "custom": {
                    "mode": "non-dbfred",
                    "title": {
                        "type": "title",
                        "title": "\"Latest Updates on EU Regulations\""
                    }
                }
            }
        }
    },
    "hasLivechat": false
}

Usage Example

from captivate_ai_api import Captivate, TextMessageModel



@app.post("/chat")
async def handle_chat(data: any):
    try:
        # Create Captivate instance using the request data
        captivate = Captivate(**data.dict())
        captivate.set_conversation_title('Lord of the rings')

        # Prepare messages
        response_messages = [
            TextMessageModel(text="Welcome to our platform!"),
            ButtonMessageModel(buttons={"title": "Learn More", "options": [{"label":"Yes","value":"Yes"}]}),
            TableMessageModel(table="<table><tr><th>Name</th><th>Age</th></tr><tr><td>Alice</td><td>30</td></tr></table>"),
            CardCollectionModel(cards=[CardMessageModel(
                text="Special Offer",
                description="Get 20% off your next purchase.",
                image_url="https://example.com/offer.png",
                link="https://example.com/deals"
            )]),
            HtmlMessageModel(html="<h2>Today's Highlights</h2><ul><li>News Item 1</li><li>News Item 2</li></ul>"),
            FileCollectionModel(files=[FileModel(type='application/pdf',url="https://example.com/manual.pdf", filename="UserManual.pdf")] ),
            {"type": "custom", "content": "This is a custom message."}
        ]
        
        # Set the response messages
        captivate.set_response(response_messages)

        # Outgoing actions Both 'payload' & 'data' works for backwards compatibliity. Moving forward it is recommended to use 'data'
        outgoing_actions = [
            ActionModel(id="navigate", payload={"url": "https://example.com"}),
            ActionModel(id="submit", data={"form_id": "1234"})
        ] 
        captivate.set_outgoing_action(outgoing_actions)

        return captivate.get_response() #Returns data to captivate platform in the correct format

Expected Response from /chat Endpoint

When you send the POST request to the /chat endpoint, the response will look as follows:

{
    "response": [
        {
            "type": "text",
            "text": "Welcome to our platform!"
        },
        {
            "type": "button",
            "buttons": {
                "title": "Learn More",
                "options": {
                    "label":"Yes",
                    "value":"Yes"
                }
            }
        },
        {
            "type": "table",
            "table": "<table><tr><th>Name</th><th>Age</th></tr><tr><td>Alice</td><td>30</td></tr></table>"
        },
        {
            "type": "cards",
            "text": "Special Offer",
            "description": "Get 20% off your next purchase.",
            "image_url": "https://example.com/offer.png",
            "link": "https://example.com/deals"
        },
        {
            "type": "html",
            "html": "<h2>Today's Highlights</h2><ul><li>News Item 1</li><li>News Item 2</li></ul>"
        },
        { 
        "type":"files",
        "title":"these are the files",
        "files":[{
            "type": "application/pdf",
            "url": "https://example.com/manual.pdf",
            "filename": "UserManual.pdf"
            }]
        },
        {
            "type": "alert",
            "RootModel": {
                "priority": "high",
                "message": "System maintenance scheduled."
            }
        }
    ],
    "session_id": "lance_catcher_test_69c35e3e-7ff4-484e-8e36-792a62567b79",
    "metadata": {
        "internal": {
            "channelMetadata": {
                "user": {
                    "firstName": "Lance",
                    "lastName": "safa",
                    "email": "[e-mail]"
                },
                "channelMetadata": {
                    "channel": "custom-channel",
                    "channelData": {}
                },
                "custom": {
                    "mode": "non-dbfred",
                    "title": {
                        "type": "title",
                        "title": "Lord of the rings"
                    }
                },
                "conversationCreatedAt": null,
                "conversationUpdatedAt": null
            }
        }
    },
    "outgoing_action": [
        {
            "id": "navigate",
            "payload": {
                "url": "https://example.com"
            },
            "data": {
                "url": "https://example.com"
            }
        },
        {
            "id": "submit",
            "payload": {
                "form_id": "1234"
            },
            "data": {
                "form_id": "1234"
            }
        }
    ],
    "hasLivechat": false
}

Functions Overview

1. get_session_id

def get_session_id(self) -> str:

• Description: Returns the value of session_id.
• Example:

session_id = captivate_instance.get_session_id()

2. get_user_input

def get_user_input(self) -> Optional[str]:

• Description: Returns the value of user_input.
• Example:

user_input = captivate_instance.get_user_input()

3. set_conversation_title

def set_conversation_title(self, title: str):

• Description: Sets the conversation title in the custom metadata.
• Example:

captivate_instance.set_conversation_title("New Conversation Title")

4. get_conversation_title

def get_conversation_title(self) -> Optional[str]:

• Description: Retrieves the conversation title from the custom metadata.
• Example:

conversation_title = captivate_instance.get_conversation_title()

5. set_metadata

def set_metadata(self, key: str, value: Any):

• Description: Sets a key-value pair in the custom metadata.
• Example:

captivate_instance.set_metadata("custom_key", "custom_value")

6. get_metadata

def get_metadata(self, key: str) -> Optional[Any]:

• Description: Retrieves the value for a given key in the custom metadata.
• Example:

metadata_value = captivate_instance.get_metadata("custom_key")

7. remove_metadata

def remove_metadata(self, key: str) -> bool:

• Description: Removes a key from the custom metadata.
• Example:

captivate_instance.remove_metadata("custom_key")

8. get_channel

def get_channel(self) -> Optional[str]:

• Description: Retrieves the channel from the metadata.
• Example:

channel = captivate_instance.get_channel()

9. get_user

def get_user(self) -> Optional[UserModel]:

• Description: Retrieves the user from the metadata.
• Example:

user = captivate_instance.get_user()

10. set_user

def set_user(self, user: UserModel) -> None:

• Description: Sets the user in the metadata.
• Example:

captivate_instance.set_user(UserModel(firstName="John", lastName="Doe"))

11. get_created_at

def get_created_at(self) -> Optional[str]:

• Description: Returns the conversationCreatedAt timestamp from the metadata.
• Example:

created_at = captivate_instance.get_created_at()

12. get_updated_at

def get_updated_at(self) -> Optional[str]:

• Description: Returns the conversationUpdatedAt timestamp from the metadata.
• Example:

updated_at = captivate_instance.get_updated_at()

13. get_has_livechat

def get_has_livechat(self) -> bool:

• Description: Returns the value of hasLivechat.
• Example:

has_livechat = captivate_instance.get_has_livechat()

14. set_response

def set_response(self, response: List[Union[TextMessageModel, FileCollectionModel, ButtonMessageModel, TableMessageModel, CardCollectionModel, HtmlMessageModel, dict]]) -> None:

• Description: Sets the response messages in the Captivate instance.
• Example:

captivate_instance.set_response([
            TextMessageModel(text="Welcome to our platform!"),
            ButtonMessageModel(buttons={"title": "Learn More", "action": "navigate"}),
            TableMessageModel(table="<table><tr><th>Name</th><th>Age</th></tr><tr><td>Alice</td><td>30</td></tr></table>"),
            CardCollectionModel(cards=[CardMessageModel(
                text="Special Offer",
                description="Get 20% off your next purchase.",
                image_url="https://example.com/offer.png",
                link="https://example.com/deals"
            )]),
            HtmlMessageModel(html="<h2>Today's Highlights</h2><ul><li>News Item 1</li><li>News Item 2</li></ul>"),
           FileCollectionModel(title="See files below", files=[FileModel(type='application/pdf',url="https://example.com/manual.pdf", filename="UserManual.pdf")] ),
            {"type": "custom", "content": "This is a custom message."}
            ])

15. get_incoming_action

def get_incoming_action(self) -> Optional[List[ActionModel]]:

• Description: Retrieves the incoming actions from the response object, if present.
• Example:

incoming_actions = captivate_instance.get_incoming_action()

16. set_outgoing_action

def set_outgoing_action(self, actions: List[ActionModel]) -> None:

• Description: Sets the outgoing actions in the response object.
• Example:

captivate_instance.set_outgoing_action([
    ActionModel(id="navigate", data={"url": "https://example.com"})
])

17. get_response

def get_response(self) -> Optional[str]:

• Description: Returns the CaptivateResponseModel as a JSON string if it exists, otherwise returns None.
• Example:

response_json = captivate_instance.get_response()

18. async_send_message

async def async_send_message(self, environment: str = "dev") -> Dict[str, Any]:

• Description: The async_send_message method is an asynchronous function that sends the conversation data (including messages and actions) to the captivate async messsage API endpoint, depending on the environment (dev or prod)
• Example:

# Create an instance of Captivate
captivate = Captivate(session_id="12345", hasLivechat=True, metadata=metadata)

# Set a message and actions
captivate.set_response([TextMessageModel(text="Hello, World!")])

# Send the message to the API in 'dev' environment
response = await captivate.async_send_message(environment="dev")

18. download_file_to_memory

 async def download_file_to_memory(self, file_info: Dict[str, Any]) -> io.BytesIO:

• Description: Downloads a file from the given dictionary and stores it in memory.

• Example:

captivate_instance.download_file_to_memory(file_info)