lionagi 0.0.115__py3-none-any.whl → 0.0.204__py3-none-any.whl

lionagi/core/branch/branch_manager.py

@@ -0,0 +1,139 @@
+from typing import Dict, Any
+from collections import deque
+from enum import Enum
+
+
+class RequestTitle(str, Enum):
+    """
+    An enumeration of valid request titles for communication requests.
+
+    Attributes:
+        MESSAGES (str): Represents a request carrying messages.
+        TOOL (str): Represents a request for tool information or actions.
+        SERVICE (str): Represents a request related to services.
+        LLMCONFIG (str): Represents a request to configure or modify LLM settings.
+    """
+    MESSAGES = 'messages'
+    TOOL = 'tool'
+    SERVICE = 'service'
+    LLMCONFIG = 'llmconfig'
+
+
+class Request:
+    """
+    Represents a request for communication between components in the system.
+
+    Args:
+        from_name (str): The name of the sender.
+        to_name (str): The name of the recipient.
+        title (Union[str, RequestTitle]): The title of the request, indicating its type or category.
+        request (Any): The actual content or data of the request.
+
+    Raises:
+        ValueError: If the request title is invalid or not recognized.
+    """
+
+    def __init__(self, from_name, to_name, title, request):
+        self.from_ = from_name
+        self.to_ = to_name
+        try:
+            if isinstance(title, str):
+                title = RequestTitle(title)
+            if isinstance(title, RequestTitle):
+                self.title = title
+            else:
+                raise ValueError(f'Invalid request title. Valid titles are {list(RequestTitle)}')
+        except:
+            raise ValueError(f'Invalid request title. Valid titles are {list(RequestTitle)}')
+        self.request = request
+
+
+class BranchManager:
+    """
+    Manages branches and their communication requests within a system.
+
+    Args:
+        sources (Dict[str, Any]): Initial mapping of source names to their respective details.
+
+    Methods:
+        add_source: Adds a new source to the manager.
+        delete_source: Removes a source from the manager.
+        collect: Collects outgoing requests from a specified source and queues them for the recipient.
+        send: Sends the queued requests to their respective recipients in the system.
+    """
+
+    def __init__(self, sources: Dict[str, Any]):
+        self.sources = sources
+        self.requests = {}
+        for key in self.sources.keys():
+            self.requests[key] = {}
+
+    def add_source(self, sources: Dict[str, Any]):
+        """
+        Adds a new source or multiple sources to the manager.
+
+        Args:
+            sources (Dict[str, Any]): A dictionary mapping new source names to their details.
+
+        Raises:
+            ValueError: If any of the provided source names already exist.
+        """
+        for key in sources.keys():
+            if key in self.sources:
+                raise ValueError(f'{key} exists, please input a different name.')
+            self.sources[key] = {}
+
+    def delete_source(self, source_name):
+        """
+        Deletes a source from the manager by name.
+
+        Args:
+            source_name (str): The name of the source to delete.
+
+        Raises:
+            ValueError: If the specified source name does not exist.
+        """
+        if source_name not in self.sources:
+            raise ValueError(f'{source_name} does not exist.')
+        self.sources.pop(source_name)
+
+    def collect(self, from_name):
+        """
+        Collects all outgoing requests from a specified source.
+
+        Args:
+            from_name (str): The name of the source from which to collect outgoing requests.
+
+        Raises:
+            ValueError: If the specified source name does not exist.
+        """
+        if from_name not in self.sources:
+            raise ValueError(f'{from_name} does not exist.')
+        while self.sources[from_name].pending_outs:
+            request = self.sources[from_name].pending_outs.popleft()
+            if request.from_ not in self.requests[request.to_]:
+                self.requests[request.to_] = {request.from_: deque()}
+            self.requests[request.to_][request.from_].append(request)
+
+    def send(self, to_name):
+        """
+        Sends all queued requests to a specified recipient.
+
+        Args:
+            to_name (str): The name of the recipient to whom the requests should be sent.
+
+        Raises:
+            ValueError: If the specified recipient name does not exist or there are no requests to send.
+        """
+        if to_name not in self.sources:
+            raise ValueError(f'{to_name} does not exist.')
+        if not self.requests[to_name]:
+            return
+        else:
+            for key in list(self.requests[to_name].keys()):
+                request = self.requests[to_name].pop(key)
+                if key not in self.sources[to_name].pending_ins:
+                    self.sources[to_name].pending_ins[key] = request
+                else:
+                    self.sources[to_name].pending_ins[key].append(request)
+

lionagi/core/branch/conversation.py

@@ -0,0 +1,484 @@
+import json
+import pandas as pd
+from datetime import datetime
+from typing import Any, Optional, Dict, Union
+from lionagi.schema import DataLogger
+from lionagi.utils import lcall, as_dict
+from ..messages.messages import Message, System, Instruction, Response
+from ..core_util import sign_message, validate_messages
+
+
+class Conversation:
+    """
+    Represents a conversation with messages, tools, and instructions.
+
+    A `Conversation` is a container for messages exchanged in a conversation, as well as tools and instructions
+    for interacting with external services or tools.
+
+    Attributes:
+        messages (pd.DataFrame): A DataFrame containing conversation messages.
+        _logger (DataLogger): An instance of DataLogger for logging.
+    """
+
+    def __init__(self, dir: Optional[str] = None) -> None:
+        """
+        Initializes a Conversation object with an empty DataFrame for messages and a DataLogger.
+
+        Args:
+            dir (Optional[str]): The directory path for storing logs. Defaults to None.
+
+        Examples:
+            >>> conversation = Conversation(dir='logs/')
+        """
+        self.messages = pd.DataFrame(columns=["node_id", "role", "sender", "timestamp", "content"])
+        self.logger = DataLogger(dir=dir)
+
+    def _create_message(
+        self,
+        system: Optional[Union[dict, list, System]] = None,
+        instruction: Optional[Union[dict, list, Instruction]] = None,
+        context: Optional[Union[str, Dict[str, Any]]] = None,
+        response: Optional[Union[dict, list, Response]] = None,
+        sender: Optional[str] = None
+    ) -> Message:
+        """
+        Creates a Message object based on the given parameters.
+
+        Only one of `system`, `instruction`, or `response` can be provided to create a message.
+
+        Args:
+            system (Optional[Union[dict, list, System]]): The system message content.
+            instruction (Optional[Union[dict, list, Instruction]]): The instruction content.
+            context (Optional[Union[str, Dict[str, Any]]]): The context associated with the instruction.
+            response (Optional[Union[dict, list, Response]]): The response content.
+            sender (Optional[str]): The sender of the message.
+
+        Returns:
+            Message: A message object created from the provided parameters.
+
+        Raises:
+            ValueError: If more than one or none of the parameters (system, instruction, response) are provided.
+
+        Examples:
+            >>> conversation = Conversation()
+            >>> msg = conversation._create_message(system="System message", sender="system")
+        """
+        
+        if sum(lcall([system, instruction, response], bool)) != 1:
+            raise ValueError("Error: Message must have one and only one role.")
+        else:
+            if isinstance(any([system, instruction, response]), Message):
+                if system:
+                    return system
+                elif instruction:
+                    return instruction
+                elif response:
+                    return response
+
+            msg = 0
+            if response:
+                msg = Response(response=response, sender=sender)
+            elif instruction:
+                msg = Instruction(instruction=instruction, 
+                                  context=context, sender=sender)
+            elif system:
+                msg = System(system=system, sender=sender)
+            return msg
+        
+    def add_message(
+        self,
+        system: Optional[Union[dict, list, System]] = None,
+        instruction: Optional[Union[dict, list, Instruction]] = None,
+        context: Optional[Union[str, Dict[str, Any]]] = None,
+        response: Optional[Union[dict, list, Response]] = None,
+        sender: Optional[str] = None
+    ) -> None:
+        """
+        Creates and adds a new message to the conversation's messages DataFrame.
+        Args:
+            system (Optional[System]): Content for a system message.
+            instruction (Optional[Instruction]): Content for an instruction message.
+            context (Optional[Union[str, Dict[str, Any]]]): Context for the instruction message.
+            response (Optional[Response]): Content for a response message.
+            sender (Optional[str]): The sender of the message.
+
+        Examples:
+            >>> conversation = Conversation()
+            >>> conversation.add_message(instruction="What's the weather?", sender="user")
+        """
+        msg = self._create_message(
+            system=system, instruction=instruction, 
+            context=context, response=response, sender=sender
+        )
+        message_dict = msg.to_dict()
+        if isinstance(as_dict(message_dict['content']), dict):
+            message_dict['content'] = json.dumps(message_dict['content'])
+        message_dict['timestamp'] = datetime.now()
+        self.messages.loc[len(self.messages)] = message_dict
+    
+    @property
+    def last_row(self) -> pd.Series:
+        """
+        Retrieves the last row from the messages DataFrame.
+
+        Returns:
+            pd.Series: A Series object representing the last message.
+        """
+        return self.messages.iloc[-1]
+    
+    @property
+    def first_system(self) -> pd.Series:
+        """
+        Retrieves the first system message from the messages DataFrame.
+
+        Returns:
+            pd.Series: A Series object representing the first system message.
+        """
+        return self.messages[self.messages.role == 'system'].iloc[0]
+        
+    @property
+    def last_response(self) -> pd.Series:
+        """
+        Retrieves the last response message from the messages DataFrame.
+
+        Returns:
+            pd.Series: A Series object representing the last response message.
+        """
+        return self.get_last_rows(role='assistant')
+    
+    @property
+    def last_instruction(self) -> pd.Series:
+        """
+        Retrieves the last instruction message from the messages DataFrame.
+
+        Returns:
+            pd.Series: A Series object representing the last instruction message.
+        """
+        return self.get_last_rows(role='user')
+    
+    def get_last_rows(
+        self, 
+        sender: Optional[str] = None, 
+        role: Optional[str] = None, 
+        n: int = 1, 
+        sign_ = False
+    ) -> Union[pd.DataFrame, pd.Series]:
+        """
+        Retrieves the last n rows from the messages DataFrame filtered by sender or role.
+
+        Args:
+            sender (Optional[str]): The sender filter for the messages.
+            role (Optional[str]): The role filter for the messages.
+            n (int): The number of rows to retrieve.
+            sign_: If sign messages with a sender identifier.
+
+        Returns:
+            Union[pd.DataFrame, pd.Series]: The last n messages as a DataFrame or a single message as a Series.
+
+        Raises:
+            ValueError: If both sender and role are provided or if none is provided.
+        """
+
+        if sender is None and role is None:
+            outs = self.messages.iloc[-n:]
+        elif sender and role:
+            outs = self.messages[(self.messages['sender'] == sender) & (self.messages['role'] == role)].iloc[-n:]
+        elif sender:
+            outs = self.messages[self.messages['sender'] == sender].iloc[-n:]
+        else:
+            outs = self.messages[self.messages['role'] == role].iloc[-n:]
+
+        return sign_message(outs, sender) if sign_ else outs
+
+    def filter_messages_by(
+        self,
+        role: Optional[str] = None, 
+        sender: Optional[str] = None,
+        start_time: Optional[datetime] = None,
+        end_time: Optional[datetime] = None,
+        content_keywords: Optional[Union[str, list]] = None,
+        case_sensitive: bool = False
+    ) -> pd.DataFrame:
+        """
+        Retrieves messages filtered by a specific criterion.
+
+        Args:
+            role (Optional[str]): The role to filter the messages.
+            sender (Optional[str]): The sender to filter the messages.
+            start_time (Optional[datetime]): The start time to filter the messages.
+            end_time (Optional[datetime]): The end time to filter the messages.
+            content_keywords (Optional[Union[str, list]]): The content to filter the messages.
+            case_sensitive (bool): Flag to indicate if the search should be case sensitive. Defaults to False.
+
+        Returns:
+            pd.DataFrame: A DataFrame containing filtered messages.
+
+        Raises:
+            ValueError: If more than one or none of the filtering criteria are provided.
+        """
+        outs = self.messages.copy()
+        
+        if content_keywords:
+            outs = self.search_keywords(content_keywords, case_sensitive)
+        
+        outs = outs[outs['role'] == role] if role else outs
+        outs = outs[outs['sender'] == sender] if sender else outs
+        outs = outs[outs['timestamp'] > start_time] if start_time else outs
+        outs = outs[outs['timestamp'] < end_time] if end_time else outs
+        return outs
+
+    def replace_keyword(
+        self, 
+        keyword: str, 
+        replacement: str, 
+        case_sensitive: bool = False
+    ) -> None:
+        """
+        Replaces a keyword in the content of all messages with a replacement string.
+
+        Args:
+            keyword (str): The keyword to replace.
+            replacement (str): The string to replace the keyword with.
+            case_sensitive (bool, optional): Flag to indicate if the replacement should be case sensitive. Defaults to False.
+        """
+        if not case_sensitive:
+            self.messages["content"] = self.messages["content"].str.replace(
+                keyword, replacement, case=False
+            )
+        else:
+            self.messages["content"] = self.messages["content"].str.replace(
+                keyword, replacement
+            )
+
+    def search_keywords(
+        self, 
+        keywords: Union[str, list],
+        case_sensitive: bool = False
+    ) -> pd.DataFrame:
+        """
+        Searches for a keyword in the content of all messages and returns the messages containing it.
+
+        Args:
+            keywords (str): The keywords to search for.
+            case_sensitive (bool, optional): Flag to indicate if the search should be case sensitive. Defaults to False.
+
+        Returns:
+            pd.DataFrame: A DataFrame containing messages with the specified keyword.
+        """
+        if isinstance(keywords, list):
+            keywords = '|'.join(keywords)
+        if not case_sensitive:
+            return self.messages[
+                self.messages["content"].str.contains(keywords, case=False)
+            ]
+        return self.messages[self.messages["content"].str.contains(keywords)]
+
+    def remove_from_messages(self, message_id: str) -> bool:
+        """
+        Removes a message from the conversation based on its message ID.
+
+        Args:
+            message_id (str): The ID of the message to be removed.
+
+        Returns:
+            bool: True if the message was successfully removed, False otherwise.
+        """
+        initial_length = len(self.messages)
+        self.messages = self.messages[self.messages["node_id"] != message_id]
+        return len(self.messages) < initial_length
+
+    def update_messages_content(
+        self, 
+        message_id: str, 
+        col: str, 
+        value: Any
+    ) -> bool:
+        """
+        Updates the content of a specific message in the conversation.
+
+        Args:
+            message_id (str): The ID of the message to be updated.
+            col (str): The column of the message that needs to be updated.
+            value (Any): The new value to be set for the specified column.
+
+        Returns:
+            bool: True if the update was successful, False otherwise.
+
+        Examples:
+            >>> conversation = Conversation()
+            >>> conversation.add_message(system="Initial message", sender="system")
+            >>> success = conversation.update_messages_content(
+            ...     message_id="1", col="content", value="Updated message")
+        """
+        index = self.messages.index[self.messages["id_"] == message_id].tolist()
+        if index:
+            self.messages.at[index[0], col] = value
+            return True
+        return False
+
+    def info(self, use_sender: bool = False) -> Dict[str, int]:
+        """
+        Provides a summary of the conversation messages.
+
+        Args:
+            use_sender (bool, optional): Determines whether to summarize by sender or by role. Defaults to False.
+
+        Returns:
+            Dict[str, int]: A dictionary containing counts of messages either by role or sender.
+        """
+        messages = self.messages['sender'] if use_sender else self.messages['role']
+        result = messages.value_counts().to_dict()
+        result['total'] = len(self.messages)
+        return result
+
+    @property
+    def describe(self) -> Dict[str, Any]:
+        """
+        Describes the conversation with various statistics and information.
+
+        Returns:
+            Dict[str, Any]: A dictionary containing information such as total number of messages, summary by role,
+                            and individual messages.
+        """
+        return {
+            "total_messages": len(self.messages),
+            "summary_by_role": self.info(),
+            "messages": [
+                msg.to_dict() for _, msg in self.messages.iterrows()
+            ],
+        }
+
+    def history(
+        self, begin_: Optional[datetime] = None, end_: Optional[datetime] = None
+    ) -> pd.DataFrame:
+        """
+        Retrieves a history of messages within a specified date range.
+
+        Args:
+            begin_ (Optional[datetime], optional): The start date of the message history. Defaults to None.
+            end_ (Optional[datetime], optional): The end date of the message history. Defaults to None.
+
+        Returns:
+            pd.DataFrame: A DataFrame containing messages within the specified date range.
+        """
+        
+        if isinstance(begin_, str):
+            begin_ = datetime.strptime(begin_, '%Y-%m-%d')
+        if isinstance(end_, str):
+            end_ = datetime.strptime(end_, '%Y-%m-%d')
+        if begin_ and end_:
+            return self.messages[
+                (self.messages["timestamp"].dt.date >= begin_.date())
+                & (self.messages["timestamp"].dt.date <= end_.date())
+            ]
+        elif begin_:
+            return self.messages[(self.messages["timestamp"].dt.date >= begin_.date())]
+        elif end_:
+            return self.messages[(self.messages["timestamp"].dt.date <= end_.date())]
+        return self.messages
+
+    def clone(self) -> 'Conversation':
+        """
+        Creates a clone of the current conversation.
+
+        Returns:
+            Conversation: A new Conversation object that is a clone of the current conversation.
+        """
+        cloned = Conversation()
+        cloned.logger.set_dir(self.logger.dir)
+        cloned.messages = self.messages.copy()
+        return cloned
+
+    # def merge_conversation(self, other: 'Conversation', update: bool = False,) -> None:
+    #     """
+    #     Merges another conversation into the current one.
+    #
+    #     Args:
+    #         other (Conversation): The other conversation to merge with the current one.
+    #         update (bool, optional): If True, updates the first system message before merging. Defaults to False.
+    #     """
+    #     if update:
+    #         self.first_system = other.first_system.copy()
+    #     df = pd.concat([self.messages.copy(), other.messages.copy()], ignore_index=True)
+    #     self.messages = df.drop_duplicates().reset_index(drop=True, inplace=True)
+
+    def rollback(self, steps: int) -> None:
+        """
+        Rollbacks the conversation by a specified number of steps (messages).
+
+        Args:
+            steps (int): The number of steps to rollback.
+
+        Raises:
+            ValueError: If steps are not a non-negative integer or greater than the number of messages.
+        """
+        if steps < 0 or steps > len(self.messages):
+            raise ValueError("Steps must be a non-negative integer less than or equal to the number of messages.")
+        self.messages = self.messages[:-steps].reset_index(drop=True)
+
+    def reset(self) -> None:
+        """
+        Resets the conversation, clearing all messages.
+        """
+        self.messages = pd.DataFrame(columns=self.messages.columns)
+
+    def to_csv(self, filepath: str, **kwargs) -> None:
+        """
+        Exports the conversation messages to a CSV file.
+
+        Args:
+            filepath (str): The file path where the CSV will be saved.
+            **kwargs: Additional keyword arguments for `pandas.DataFrame.to_csv` method.
+        """
+        self.messages.to_csv(filepath, **kwargs)
+
+    def from_csv(self, filepath: str, **kwargs) -> None:
+        """
+        Imports conversation messages from a CSV file.
+
+        Args:
+            filepath (str): The file path of the CSV to be read.
+            **kwargs: Additional keyword arguments for `pandas.read_csv` method.
+        """
+        self.messages = pd.read_csv(filepath, **kwargs)
+
+    def to_json(self, filepath: str) -> None:
+        """
+        Exports the conversation messages to a JSON file.
+
+        Args:
+            filepath (str): The file path where the JSON will be saved.
+        """
+        self.messages.to_json(
+            filepath, orient="records", lines=True, date_format="iso")
+
+    def from_json(self, filepath: str) -> None:
+        """
+        Imports conversation messages from a JSON file.
+
+        Args:
+            filepath (str): The file path of the JSON to be read.
+        """
+        self.reset()
+        self.messages = pd.read_json(filepath, orient="records", lines=True)
+
+    def extend(self, messages: pd.DataFrame, **kwargs) -> None:
+        """
+        Extends the current conversation messages with additional messages from a DataFrame.
+
+        Args:
+            messages (pd.DataFrame): The DataFrame containing messages to be added to the conversation.
+            kwargs: for pd.df.drop_duplicates
+        """
+        
+        validate_messages(messages)
+        try:
+            if len(messages.dropna(how='all')) > 0 and len(self.messages.dropna(how='all')) > 0:
+                self.messages = pd.concat([self.messages, messages], ignore_index=True)
+                self.messages.drop_duplicates(
+                    inplace=True, subset=['node_id'], keep='first', **kwargs
+                )
+                self.messages.reset_index(drop=True, inplace=True)
+                return
+        except Exception as e:
+            raise ValueError(f"Error in extending messages: {e}")

lionagi/core/core_util.py

@@ -0,0 +1,59 @@
+import json
+from ..utils.sys_util import strip_lower
+
+
+def sign_message(messages, sender: str):
+    """
+    Sign messages with a sender identifier.
+
+    Args:
+        messages (pd.DataFrame): A DataFrame containing messages with columns 'node_id', 'role', 'sender', 'timestamp', and 'content'.
+        sender (str): The sender identifier to be added to the messages.
+
+    Returns:
+        pd.DataFrame: A new DataFrame with the sender identifier added to each message.
+
+    Raises:
+        ValueError: If the 'sender' is None or 'None'.
+    """
+    if sender is None or strip_lower(sender) == 'none':
+        raise ValueError("sender cannot be None")
+    df = messages.copy()
+
+    for i in df.index:
+        if not df.loc[i, 'content'].startswith('Sender'):
+            df.loc[i, 'content'] = f"Sender {sender}: {df.loc[i, 'content']}"
+        else:
+            content = df.loc[i, 'content'].split(':', 1)[1]
+            df.loc[i, 'content'] = f"Sender {sender}: {content}"
+    return df
+
+
+def validate_messages(messages):
+    """
+    Validate the structure and content of a messages DataFrame.
+
+    Args:
+        messages (pd.DataFrame): A DataFrame containing messages with columns 'node_id', 'role', 'sender', 'timestamp', and 'content'.
+
+    Returns:
+        bool: True if the messages DataFrame is valid; otherwise, raises a ValueError.
+
+    Raises:
+        ValueError: If the DataFrame structure is invalid or if it contains null values, roles other than ["system", "user", "assistant"],
+                    or content that cannot be parsed as JSON strings.
+    """
+    if list(messages.columns) != ['node_id', 'role', 'sender', 'timestamp', 'content']:
+        raise ValueError('Invalid messages dataframe. Unmatched columns.')
+    if messages.isnull().values.any():
+        raise ValueError('Invalid messages dataframe. Cannot have null.')
+    if not all(role in ['system', 'user', 'assistant'] for role in messages['role'].unique()):
+        raise ValueError('Invalid messages dataframe. Cannot have role other than ["system", "user", "assistant"].')
+    for cont in messages['content']:
+        if cont.startswith('Sender'):
+            cont = cont.split(':', 1)[1]
+        try:
+            json.loads(cont)
+        except:
+            raise ValueError('Invalid messages dataframe. Content expect json string.')
+    return True