mb-rag 1.0.123__py3-none-any.whl → 1.0.125__py3-none-any.whl

mb_rag/chatbot/basic.py

@@ -220,13 +220,14 @@ class ConversationModel:
         self.chatbot = ModelFactory(model_type, model_name, **kwargs)
 
     def initialize_conversation(self,
-                               file_path: Optional[str],
-                               question: Optional[str],
-                               context: Optional[str]) -> None:
+                                question: Optional[str],
+                                context: Optional[str] = None,
+                                file_path: Optional[str]=None) -> None:
         """Initialize conversation state"""
         if file_path:
-            self.file_path = file_path
+            self.file_path = file_path  
             self.load_conversation(file_path)
+                
         else:
             if not question:
                 raise ValueError("Question is required.")
@@ -371,7 +372,7 @@ class ConversationModel:
             raise ValueError(f"Error loading conversation from s3: {e}")
 
     def _load_from_file(self, file_path: str) -> List[Any]:
-        """Load conversation from file"""
+        """Load conversation from file"""            
         try:
             with open(file_path, 'r') as f:
                 lines = f.readlines()

mb_rag/rag/embeddings.py

@@ -1,480 +1,729 @@
-## Function to generate embeddings for the RAG model
+"""
+RAG (Retrieval-Augmented Generation) Embeddings Module
+
+This module provides functionality for generating and managing embeddings for RAG models.
+It supports multiple embedding models (OpenAI, Ollama, Google, Anthropic) and includes
+features for text processing, embedding generation, vector store management, and 
+conversation handling.
+
+Example Usage:
+    ```python
+    # Initialize embedding generator
+    em_gen = embedding_generator(
+        model="openai",
+        model_type="text-embedding-3-small",
+        vector_store_type="chroma"
+    )
+
+    # Generate embeddings from text
+    em_gen.generate_text_embeddings(
+        text_data_path=['./data/text.txt'],
+        chunk_size=500,
+        chunk_overlap=5,
+        folder_save_path='./embeddings'
+    )
+
+    # Load embeddings and create retriever
+    em_loading = em_gen.load_embeddings('./embeddings')
+    em_retriever = em_gen.load_retriever(
+        './embeddings',
+        search_params=[{"k": 2, "score_threshold": 0.1}]
+    )
+
+    # Query embeddings
+    results = em_retriever.invoke("What is the text about?")
+
+    # Generate RAG chain for conversation
+    rag_chain = em_gen.generate_rag_chain(retriever=em_retriever)
+    response = em_gen.conversation_chain("Tell me more", rag_chain)
+    ```
+
+Features:
+    - Multiple model support (OpenAI, Ollama, Google, Anthropic)
+    - Text processing and chunking
+    - Embedding generation and storage
+    - Vector store management
+    - Retrieval operations
+    - Conversation chains
+    - Web crawling integration
+
+Classes:
+    - ModelProvider: Base class for model loading and validation
+    - TextProcessor: Handles text processing operations
+    - embedding_generator: Main class for RAG operations
+"""
 
 import os
 import shutil
 import importlib.util
+from typing import List, Dict, Optional, Union, Any
 from langchain.text_splitter import (
     CharacterTextSplitter,
     RecursiveCharacterTextSplitter,
     SentenceTransformersTokenTextSplitter,
     TokenTextSplitter)
-from langchain_community.document_loaders import TextLoader,FireCrawlLoader
+from langchain_community.document_loaders import TextLoader, FireCrawlLoader
 from langchain_community.vectorstores import Chroma
-from ..utils.extra  import load_env_file
+from ..utils.extra import load_env_file
 from langchain.chains import create_history_aware_retriever, create_retrieval_chain
 from langchain.chains.combine_documents import create_stuff_documents_chain
 from langchain_core.messages import HumanMessage, SystemMessage
 from langchain_core.prompts import ChatPromptTemplate, MessagesPlaceholder
-import time
 
 load_env_file()
 
-__all__ = ['embedding_generator', 'load_rag_model']
+__all__ = ['embedding_generator', 'load_embedding_model']
 
-def check_package(package_name):
+class ModelProvider:
     """
-    Check if a package is installed
-    Args:
-        package_name (str): Name of the package
-    Returns:
-        bool: True if package is installed, False otherwise
+    Base class for managing different model providers and their loading logic.
+
+    This class provides static methods for loading different types of embedding models
+    and checking package dependencies.
+
+    Methods:
+        check_package: Check if a Python package is installed
+        get_rag_openai: Load OpenAI embedding model
+        get_rag_ollama: Load Ollama embedding model
+        get_rag_anthropic: Load Anthropic model
+        get_rag_google: Load Google embedding model
+
+    Example:
+        ```python
+        # Check if a package is installed
+        has_openai = ModelProvider.check_package("langchain_openai")
+
+        # Load an OpenAI model
+        model = ModelProvider.get_rag_openai("text-embedding-3-small")
+        ```
     """
-    return importlib.util.find_spec(package_name) is not None
-
-def get_rag_openai(model_type: str = 'text-embedding-3-small',**kwargs):
-    """
-    Load model from openai for RAG
-    Args:
-        model_type (str): Name of the model
-        **kwargs: Additional arguments (temperature, max_tokens, timeout, max_retries, api_key etc.)
-    Returns:
-        ChatOpenAI: Chatbot model
-    """
-    if not check_package("langchain_openai"):
-        raise ImportError("OpenAI package not found. Please install it using: pip install langchain-openai")
     
-    from langchain_openai import OpenAIEmbeddings
-    return OpenAIEmbeddings(model = model_type,**kwargs)
+    @staticmethod
+    def check_package(package_name: str) -> bool:
+        """
+        Check if a Python package is installed.
 
-def get_rag_ollama(model_type: str = 'llama3',**kwargs):
-    """
-    Load model from ollama for RAG
-    Args:
-        model_type (str): Name of the model
-        **kwargs: Additional arguments (temperature, max_tokens, timeout, max_retries, api_key etc.)
-    Returns:
-        OllamaEmbeddings: Embeddings model
-    """
-    if not check_package("langchain_ollama"):
-        raise ImportError("Ollama package not found. Please install it using: pip install langchain-ollama")
-    
-    from langchain_ollama import OllamaEmbeddings
-    return OllamaEmbeddings(model = model_type,**kwargs)
+        Args:
+            package_name (str): Name of the package to check
 
-def get_rag_anthropic(model_name: str = "claude-3-opus-20240229",**kwargs):
-    """
-    Load the chatbot model from Anthropic
-    Args:
-        model_name (str): Name of the model
-        **kwargs: Additional arguments (temperature, max_tokens, timeout, max_retries, api_key etc.)
-    Returns:
-        ChatAnthropic: Chatbot model
-    """
-    if not check_package("langchain_anthropic"):
-        raise ImportError("Anthropic package not found. Please install it using: pip install langchain-anthropic")
-    
-    from langchain_anthropic import ChatAnthropic
-    kwargs["model_name"] = model_name
-    return ChatAnthropic(**kwargs)
+        """
+        return importlib.util.find_spec(package_name) is not None
 
-def get_rag_google(model_name: str = "gemini-1.5-flash",**kwargs):
-    """
-    Load the chatbot model from Google Generative AI
-    Args:
-        model_name (str): Name of the model 
-        **kwargs: Additional arguments (temperature, max_tokens, timeout, max_retries, api_key etc.)
-    Returns:
-        ChatGoogleGenerativeAI: Chatbot model
-    """
-    if not check_package("google.generativeai"):
-        raise ImportError("Google Generative AI package not found. Please install it using: pip install langchain-google-genai")
-    
-    from langchain_google_genai import GoogleGenerativeAIEmbeddings
-    kwargs["model"] = model_name
-    return GoogleGenerativeAIEmbeddings(**kwargs)
+    @staticmethod
+    def get_rag_openai(model_type: str = 'text-embedding-3-small', **kwargs):
+        """
+        Load OpenAI embedding model.
+
+        Args:
+            model_type (str): Model identifier (default: 'text-embedding-3-small')
+            **kwargs: Additional arguments for model initialization
+
+        Returns:
+            OpenAIEmbeddings: Initialized OpenAI embeddings model
+        """
+        if not ModelProvider.check_package("langchain_openai"):
+            raise ImportError("OpenAI package not found. Please install: pip install langchain-openai")
+        from langchain_openai import OpenAIEmbeddings
+        return OpenAIEmbeddings(model=model_type, **kwargs)
+
+    @staticmethod
+    def get_rag_ollama(model_type: str = 'llama3', **kwargs):
+        """
+        Load Ollama embedding model.
 
-def load_rag_model(model_name: str ='openai', model_type: str = "text-embedding-ada-002", **kwargs):
+        Args:
+            model_type (str): Model identifier (default: 'llama3')
+            **kwargs: Additional arguments for model initialization
+
+        Returns:
+            OllamaEmbeddings: Initialized Ollama embeddings model
+        """
+        if not ModelProvider.check_package("langchain_ollama"):
+            raise ImportError("Ollama package not found. Please install: pip install langchain-ollama")
+        from langchain_ollama import OllamaEmbeddings
+        return OllamaEmbeddings(model=model_type, **kwargs)
+
+    @staticmethod
+    def get_rag_anthropic(model_name: str = "claude-3-opus-20240229", **kwargs):
+        """
+        Load Anthropic model.
+
+        Args:
+            model_name (str): Model identifier (default: "claude-3-opus-20240229")
+            **kwargs: Additional arguments for model initialization
+
+        Returns:
+            ChatAnthropic: Initialized Anthropic chat model
+
+        """
+        if not ModelProvider.check_package("langchain_anthropic"):
+            raise ImportError("Anthropic package not found. Please install: pip install langchain-anthropic")
+        from langchain_anthropic import ChatAnthropic
+        kwargs["model_name"] = model_name
+        return ChatAnthropic(**kwargs)
+
+    @staticmethod
+    def get_rag_google(model_name: str = "gemini-1.5-flash", **kwargs):
+        """
+        Load Google embedding model.
+
+        Args:
+            model_name (str): Model identifier (default: "gemini-1.5-flash")
+            **kwargs: Additional arguments for model initialization
+
+        Returns:
+            GoogleGenerativeAIEmbeddings: Initialized Google embeddings model
+        """
+        if not ModelProvider.check_package("google.generativeai"):
+            raise ImportError("Google Generative AI package not found. Please install: pip install langchain-google-genai")
+        from langchain_google_genai import GoogleGenerativeAIEmbeddings
+        kwargs["model"] = model_name
+        return GoogleGenerativeAIEmbeddings(**kwargs)
+
+def load_embedding_model(model_name: str = 'openai', model_type: str = "text-embedding-ada-002", **kwargs):
     """
-    Load a RAG model from a given model name and type
+    Load a RAG model based on provider and type.
+
     Args:
-        model_name (str): Name of the model. Default is openai.
-        model_type (str): Type of the model. Default is text-embedding-ada-002.
-        **kwargs: Additional arguments (temperature, max_tokens, timeout, max_retries, api_key etc.)
+        model_name (str): Name of the model provider (default: 'openai')
+        model_type (str): Type/identifier of the model (default: "text-embedding-ada-002")
+        **kwargs: Additional arguments for model initialization
+
     Returns:
-        RAGModel: RAG model
+        Any: Initialized model instance
+
+    Example:
+        ```python
+        model = load_embedding_model('openai', 'text-embedding-3-small')
+        ```
     """
     try:
         if model_name == 'openai':
-            return get_rag_openai(model_type, **(kwargs or {}))
+            return ModelProvider.get_rag_openai(model_type, **kwargs)
         elif model_name == 'ollama':
-            return get_rag_ollama(model_type, **(kwargs or {}))
+            return ModelProvider.get_rag_ollama(model_type, **kwargs)
         elif model_name == 'google':
-            return get_rag_google(model_type, **(kwargs or {}))
+            return ModelProvider.get_rag_google(model_type, **kwargs)
         elif model_name == 'anthropic':
-            return get_rag_anthropic(model_type, **(kwargs or {}))
+            return ModelProvider.get_rag_anthropic(model_type, **kwargs)
         else:
             raise ValueError(f"Invalid model name: {model_name}")
     except ImportError as e:
         print(f"Error loading model: {str(e)}")
         return None
 
-class embedding_generator:
+class TextProcessor:
     """
-    Class to generate embeddings for the RAG model abnd chat with data
+    Handles text processing operations including file checking and tokenization.
+
+    This class provides methods for loading text files, processing them into chunks,
+    and preparing them for embedding generation.
+
     Args:
-        model: type of model. Default is openai. Options are openai, anthropic, google, ollama
-        model_type: type of model. Default is text-embedding-3-small. Options are text-embedding-3-small, text-embedding-3-large, text-embedding-ada-002 for openai.
-        vector_store_type: type of vector store. Default is chroma
-        logger: logger
-        model_kwargs: additional arguments for the model
-        vector_store_kwargs: additional arguments for the vector store
-        collection_name: name of the collection (default : test)
+        logger: Optional logger instance for logging operations
+
+    Example:
+        ```python
+        processor = TextProcessor()
+        docs = processor.tokenize(
+            ['./data.txt'],
+            'recursive_character',
+            chunk_size=1000,
+            chunk_overlap=5
+        )
+        ```
     """
-
-    def __init__(self,model: str = 'openai',model_type: str = 'text-embedding-3-small',vector_store_type:str = 'chroma' ,collection_name: str = 'test',logger= None,model_kwargs: dict = None, vector_store_kwargs: dict = None) -> None:
+    
+    def __init__(self, logger=None):
         self.logger = logger
-        self.model = load_rag_model(model_name=model, model_type=model_type, **(model_kwargs or {}))
-        if self.model is None:
-            raise ValueError(f"Failed to initialize model {model}. Please ensure required packages are installed.")
-        self.vector_store_type = vector_store_type
-        self.vector_store = self.load_vectorstore(**(vector_store_kwargs or {}))
-        self.collection_name = collection_name
 
-    def check_file(self, file_path):
-        """
-        Check if the file exists
-        """
-        if os.path.exists(file_path):
-            return True
-        else:
-            return False
+    def check_file(self, file_path: str) -> bool:
+        """Check if file exists."""
+        return os.path.exists(file_path)
 
-    def tokenize(self,text_data_path :list,text_splitter_type: str,chunk_size: int,chunk_overlap: int):
+    def tokenize(self, text_data_path: List[str], text_splitter_type: str,
+                chunk_size: int, chunk_overlap: int) -> List:
         """
-        Function to tokenize the text
+        Process and tokenize text data from files.
+
         Args:
-            text: text to tokenize
+            text_data_path (List[str]): List of paths to text files
+            text_splitter_type (str): Type of text splitter to use
+            chunk_size (int): Size of text chunks
+            chunk_overlap (int): Overlap between chunks
+
         Returns:
-            tokens
+            List: List of processed document chunks
+
         """
-        doc_data = [] 
-        for i in text_data_path:
-            if self.check_file(i):
-                text_loader = TextLoader(i)
+        doc_data = []
+        for path in text_data_path:
+            if self.check_file(path):
+                text_loader = TextLoader(path)
                 get_text = text_loader.load()
-                # print(get_text) ## testing - Need to remove
-                file_name = i.split('/')[-1]
+                file_name = path.split('/')[-1]
                 metadata = {'source': file_name}
                 if metadata is not None:
-                    for j in get_text:
-                        j.metadata = metadata
-                        doc_data.append(j)
-                if self.logger is not None:
+                    for doc in get_text:
+                        doc.metadata = metadata
+                        doc_data.append(doc)
+                if self.logger:
                     self.logger.info(f"Text data loaded from {file_name}")
             else:
-                return f"File {i} not found"
-
-        if self.logger is not None:
-            self.logger.info(f"Splitting text data into chunks of size {chunk_size} with overlap {chunk_overlap}")
-        if text_splitter_type == 'character':
-            text_splitter = CharacterTextSplitter(chunk_size=chunk_size,chunk_overlap=chunk_overlap, separator=["\n","\n\n","\n\n\n"," "])
-        if text_splitter_type == 'recursive_character':
-            text_splitter = RecursiveCharacterTextSplitter(chunk_size=chunk_size,chunk_overlap=chunk_overlap,separators=["\n","\n\n","\n\n\n"," "])
-        if text_splitter_type == 'sentence_transformers_token':
-            text_splitter = SentenceTransformersTokenTextSplitter(chunk_size=chunk_size)
-        if text_splitter_type == 'token':
-            text_splitter = TokenTextSplitter(chunk_size=chunk_size,chunk_overlap=chunk_overlap)
-        docs = text_splitter.split_documents(doc_data) 
-        if self.logger is not None:
+                return f"File {path} not found"
+
+        splitters = {
+            'character': CharacterTextSplitter(
+                chunk_size=chunk_size,
+                chunk_overlap=chunk_overlap,
+                separator=["\n", "\n\n", "\n\n\n", " "]
+            ),
+            'recursive_character': RecursiveCharacterTextSplitter(
+                chunk_size=chunk_size,
+                chunk_overlap=chunk_overlap,
+                separators=["\n", "\n\n", "\n\n\n", " "]
+            ),
+            'sentence_transformers_token': SentenceTransformersTokenTextSplitter(
+                chunk_size=chunk_size
+            ),
+            'token': TokenTextSplitter(
+                chunk_size=chunk_size,
+                chunk_overlap=chunk_overlap
+            )
+        }
+
+        if text_splitter_type not in splitters:
+            raise ValueError(f"Invalid text splitter type: {text_splitter_type}")
+
+        text_splitter = splitters[text_splitter_type]
+        docs = text_splitter.split_documents(doc_data)
+
+        if self.logger:
             self.logger.info(f"Text data splitted into {len(docs)} chunks")
         else:
             print(f"Text data splitted into {len(docs)} chunks")
-        return docs       
+        return docs
+
+class embedding_generator:
+    """
+    Main class for generating embeddings and managing RAG operations.
+
+    This class provides comprehensive functionality for generating embeddings,
+    managing vector stores, handling retrievers, and managing conversations.
+
+    Args:
+        model (str): Model provider name (default: 'openai')
+        model_type (str): Model type/identifier (default: 'text-embedding-3-small')
+        vector_store_type (str): Type of vector store (default: 'chroma')
+        collection_name (str): Name of the collection (default: 'test')
+        logger: Optional logger instance
+        model_kwargs (dict): Additional arguments for model initialization
+        vector_store_kwargs (dict): Additional arguments for vector store initialization
+
+    Example:
+        ```python
+        # Initialize generator
+        gen = embedding_generator(
+            model="openai",
+            model_type="text-embedding-3-small"
+        )
+
+        # Generate embeddings
+        gen.generate_text_embeddings(
+            text_data_path=['./data.txt'],
+            folder_save_path='./embeddings'
+        )
+
+        # Load retriever
+        retriever = gen.load_retriever('./embeddings')
+
+        # Query embeddings
+        results = gen.query_embeddings("What is this about?")
+        ```
+    """
 
-    def generate_text_embeddings(self,text_data_path: list = None,text_splitter_type: str = 'recursive_character',
-                                 chunk_size: int = 1000,chunk_overlap: int = 5,folder_save_path: str = './text_embeddings',
-                                 replace_existing: bool = False):
+    def __init__(self, model: str = 'openai', model_type: str = 'text-embedding-3-small',
+                vector_store_type: str = 'chroma', collection_name: str = 'test',
+                logger=None, model_kwargs: dict = None, vector_store_kwargs: dict = None) -> None:
+        """Initialize the embedding generator with specified configuration."""
+        self.logger = logger
+        self.model = load_embedding_model(model_name=model, model_type=model_type, **(model_kwargs or {}))
+        if self.model is None:
+            raise ValueError(f"Failed to initialize model {model}. Please ensure required packages are installed.")
+        self.vector_store_type = vector_store_type
+        self.vector_store = self.load_vectorstore(**(vector_store_kwargs or {}))
+        self.collection_name = collection_name
+        self.text_processor = TextProcessor(logger)
+
+    def check_file(self, file_path: str) -> bool:
+        """Check if file exists."""
+        return self.text_processor.check_file(file_path)
+
+    def tokenize(self, text_data_path: List[str], text_splitter_type: str,
+                chunk_size: int, chunk_overlap: int) -> List:
+        """Process and tokenize text data."""
+        return self.text_processor.tokenize(text_data_path, text_splitter_type,
+                                          chunk_size, chunk_overlap)
+
+    def generate_text_embeddings(self, text_data_path: List[str] = None,
+                               text_splitter_type: str = 'recursive_character',
+                               chunk_size: int = 1000, chunk_overlap: int = 5,
+                               folder_save_path: str = './text_embeddings',
+                               replace_existing: bool = False) -> str:
         """
-        Function to generate text embeddings
+        Generate text embeddings from input files.
+
         Args:
-            text_data_path: list of text files
-            # metadata: list of metadata for each text file. Dictionary format
-            text_splitter_type: type of text splitter. Default is recursive_character
-            chunk_size: size of the chunk
-            chunk_overlap: overlap between chunks
-            folder_save_path: path to save the embeddings
-            replace_existing: if True, replace the existing embeddings
+            text_data_path (List[str]): List of paths to text files
+            text_splitter_type (str): Type of text splitter
+            chunk_size (int): Size of text chunks
+            chunk_overlap (int): Overlap between chunks
+            folder_save_path (str): Path to save embeddings
+            replace_existing (bool): Whether to replace existing embeddings
+
         Returns:
-            None   
+            str: Status message
+
+        Example:
+            ```python
+            gen.generate_text_embeddings(
+                text_data_path=['./data.txt'],
+                folder_save_path='./embeddings'
+            )
+            ```
         """
+        if self.logger:
+            self.logger.info("Performing basic checks")
 
-        if self.logger is not None:
-            self.logger.info("Perforing basic checks")
-
-        if self.check_file(folder_save_path) and replace_existing==False:
+        if self.check_file(folder_save_path) and not replace_existing:
             return "File already exists"
         elif self.check_file(folder_save_path) and replace_existing:
-            shutil.rmtree(folder_save_path) 
+            shutil.rmtree(folder_save_path)
 
         if text_data_path is None:
             return "Please provide text data path"
 
-        assert isinstance(text_data_path, list), "text_data_path should be a list"
-        # if metadata is not None:
-        #     assert isinstance(metadata, list), "metadata should be a list"
-        #     assert len(text_data_path) == len(metadata), "Number of text files and metadata should be equal"
+        if not isinstance(text_data_path, list):
+            raise ValueError("text_data_path should be a list")
 
-        if self.logger is not None:
+        if self.logger:
             self.logger.info(f"Loading text data from {text_data_path}")
 
-        docs = self.tokenize(text_data_path,text_splitter_type,chunk_size,chunk_overlap)
+        docs = self.tokenize(text_data_path, text_splitter_type, chunk_size, chunk_overlap)
 
-        if self.logger is not None:
-            self.logger.info(f"Generating embeddings for {len(docs)} documents")    
+        if self.logger:
+            self.logger.info(f"Generating embeddings for {len(docs)} documents")
 
-        self.vector_store.from_documents(docs, self.model,collection_name=self.collection_name,persist_directory=folder_save_path)
+        self.vector_store.from_documents(docs, self.model, collection_name=self.collection_name,
+                                       persist_directory=folder_save_path)
 
-        if self.logger is not None:
+        if self.logger:
             self.logger.info(f"Embeddings generated and saved at {folder_save_path}")
 
-    def load_vectorstore(self):
-        """
-        Function to load vector store
-        Args:
-            vector_store_type: type of vector store
-        Returns:
-            vector store
-        """
+    def load_vectorstore(self, **kwargs):
+        """Load vector store."""
         if self.vector_store_type == 'chroma':
             vector_store = Chroma()
-            if self.logger is not None:
+            if self.logger:
                 self.logger.info(f"Loaded vector store {self.vector_store_type}")
             return vector_store
         else:
             return "Vector store not found"
 
-    def load_embeddings(self,embeddings_folder_path: str):
+    def load_embeddings(self, embeddings_folder_path: str):
         """
-        Function to load embeddings from the folder
+        Load embeddings from folder.
+
         Args:
-            embeddings_path: path to the embeddings
+            embeddings_folder_path (str): Path to embeddings folder
+
         Returns:
-            embeddings
+            Optional[Chroma]: Loaded vector store or None if not found
         """
         if self.check_file(embeddings_folder_path):
             if self.vector_store_type == 'chroma':
-                # embeddings_path = os.path.join(embeddings_folder_path)
-                return Chroma(persist_directory = embeddings_folder_path,embedding_function=self.model)
+                return Chroma(persist_directory=embeddings_folder_path,
+                            embedding_function=self.model)
         else:
             if self.logger:
-                self.logger.info("Embeddings file not found") 
-            return None  
-        
-    def load_retriever(self,embeddings_folder_path: str,search_type: list = ["similarity_score_threshold"],search_params: list = [{"k": 3, "score_threshold": 0.9}]):
+                self.logger.info("Embeddings file not found")
+            return None
+
+    def load_retriever(self, embeddings_folder_path: str,
+                      search_type: List[str] = ["similarity_score_threshold"],
+                      search_params: List[Dict] = [{"k": 3, "score_threshold": 0.9}]):
         """
-        Function to load retriever
+        Load retriever with search configuration.
+
         Args:
-            embeddings_path: path to the embeddings
-            search_type: list of str: type of search. Default : ["similarity_score_threshold"]
-            search_params: list of dict: parameters for the search. Default : [{"k": 3, "score_threshold": 0.9}]
+            embeddings_folder_path (str): Path to embeddings folder
+            search_type (List[str]): List of search types
+            search_params (List[Dict]): List of search parameters
+
         Returns:
-            Retriever. If multiple search types are provided, a list of retrievers is returned
+            Union[Any, List[Any]]: Single retriever or list of retrievers
+
+        Example:
+            ```python
+            retriever = gen.load_retriever(
+                './embeddings',
+                search_type=["similarity_score_threshold"],
+                search_params=[{"k": 3, "score_threshold": 0.9}]
+            )
+            ```
         """
         db = self.load_embeddings(embeddings_folder_path)
         if db is not None:
             if self.vector_store_type == 'chroma':
-                assert len(search_type) == len(search_params), "Length of search_type and search_params should be equal"
+                if len(search_type) != len(search_params):
+                    raise ValueError("Length of search_type and search_params should be equal")
                 if len(search_type) == 1:
-                    self.retriever = db.as_retriever(search_type = search_type[0],search_kwargs=search_params[0])
+                    self.retriever = db.as_retriever(search_type=search_type[0],
+                                                   search_kwargs=search_params[0])
                     if self.logger:
                         self.logger.info("Retriever loaded")
                     return self.retriever
                 else:
                     retriever_list = []
                     for i in range(len(search_type)):
-                        retriever_list.append(db.as_retriever(search_type = search_type[i],search_kwargs=search_params[i]))
+                        retriever_list.append(db.as_retriever(search_type=search_type[i],
+                                                            search_kwargs=search_params[i]))
                     if self.logger:
-                            self.logger.info("List of Retriever loaded")
+                        self.logger.info("List of Retriever loaded")
                     return retriever_list
         else:
             return "Embeddings file not found"
-        
-    def add_data(self,embeddings_folder_path: str, data: list,text_splitter_type: str = 'recursive_character',
-                 chunk_size: int = 1000,chunk_overlap: int = 5):
+
+    def add_data(self, embeddings_folder_path: str, data: List[str],
+                text_splitter_type: str = 'recursive_character',
+                chunk_size: int = 1000, chunk_overlap: int = 5):
         """
-        Function to add data to the existing db/embeddings
+        Add data to existing embeddings.
+
         Args:
-            embeddings_path: path to the embeddings
-            data: list of data to add
-            text_splitter_type: type of text splitter. Default is recursive_character
-            chunk_size: size of the chunk
-            chunk_overlap: overlap between chunks
-        Returns:
-            None
+            embeddings_folder_path (str): Path to embeddings folder
+            data (List[str]): List of text data to add
+            text_splitter_type (str): Type of text splitter
+            chunk_size (int): Size of text chunks
+            chunk_overlap (int): Overlap between chunks
         """
         if self.vector_store_type == 'chroma':
             db = self.load_embeddings(embeddings_folder_path)
             if db is not None:
-                docs = self.tokenize(data,text_splitter_type,chunk_size,chunk_overlap)
+                docs = self.tokenize(data, text_splitter_type, chunk_size, chunk_overlap)
                 db.add_documents(docs)
                 if self.logger:
                     self.logger.info("Data added to the existing db/embeddings")
-    
-    def query_embeddings(self,query: str,retriever = None):
+
+    def query_embeddings(self, query: str, retriever=None):
         """
-        Function to query embeddings
+        Query embeddings.
+
         Args:
-            search_type: type of search
-            query: query to search
+            query (str): Query string
+            retriever: Optional retriever instance
+
         Returns:
-            results
+            Any: Query results
         """
-        # if self.vector_store_type == 'chroma':
         if retriever is None:
             retriever = self.retriever
         return retriever.invoke(query)
-        # else:
-        #     return "Vector store not found"
 
-    def get_relevant_documents(self,query: str,retriever = None):
+    def get_relevant_documents(self, query: str, retriever=None):
         """
-        Function to get relevant documents
+        Get relevant documents for query.
+
         Args:
-            query: query to search
+            query (str): Query string
+            retriever: Optional retriever instance
+
         Returns:
-            results
+            List: List of relevant documents
         """
-        return self.retriever.get_relevant_documents(query)
-    
-    def generate_rag_chain(self,context_prompt: str = None,retriever = None,llm= None):
+        if retriever is None:
+            retriever = self.retriever
+        return retriever.get_relevant_documents(query)
+
+    def generate_rag_chain(self, context_prompt: str = None, retriever=None, llm=None):
         """
-        Function to start a conversation chain with a rag data. Call this to load a rag_chain module.
+        Generate RAG chain for conversation.
+
         Args:
-            context_prompt: prompt to context
-            retriever: retriever. Default is None.
-            llm: language model. Default is openai. Need chat model llm. "ChatOpenAI", "ChatAnthropic" etc. like chatbot
+            context_prompt (str): Optional context prompt
+            retriever: Optional retriever instance
+            llm: Optional language model instance
+
         Returns:
-            rag_chain_model.
+            Any: Generated RAG chain
+
+        Example:
+            ```python
+            rag_chain = gen.generate_rag_chain(retriever=retriever)
+            ```
         """
         if context_prompt is None:
-            context_prompt = ("You are an assistant for question-answering tasks. Use the following pieces of retrieved context to answer the question. "
-                              "If you don't know the answer, just say that you don't know. Use three sentences maximum and keep the answer concise. "
-                              "\n\n {context}")
-        contextualize_q_system_prompt = ("Given a chat history and the latest user question which might reference context in the chat history, formulate a standalone question which can be understood "
-                                        "without the chat history. Do NOT answer the question, just reformulate it if needed and otherwise return it as is.")
-        contextualize_q_prompt = ChatPromptTemplate.from_messages([("system", contextualize_q_system_prompt),MessagesPlaceholder("chat_history"),("human", "{input}"),])
+            context_prompt = ("You are an assistant for question-answering tasks. "
+                            "Use the following pieces of retrieved context to answer the question. "
+                            "If you don't know the answer, just say that you don't know. "
+                            "Use three sentences maximum and keep the answer concise.\n\n{context}")
+
+        contextualize_q_system_prompt = ("Given a chat history and the latest user question "
+                                       "which might reference context in the chat history, "
+                                       "formulate a standalone question which can be understood "
+                                       "without the chat history. Do NOT answer the question, "
+                                       "just reformulate it if needed and otherwise return it as is.")
+
+        contextualize_q_prompt = ChatPromptTemplate.from_messages([
+            ("system", contextualize_q_system_prompt),
+            MessagesPlaceholder("chat_history"),
+            ("human", "{input}"),
+        ])
 
         if retriever is None:
             retriever = self.retriever
         if llm is None:
-            if not check_package("langchain_openai"):
-                raise ImportError("OpenAI package not found. Please install it using: pip install langchain-openai")
+            if not ModelProvider.check_package("langchain_openai"):
+                raise ImportError("OpenAI package not found. Please install: pip install langchain-openai")
             from langchain_openai import ChatOpenAI
             llm = ChatOpenAI(model="gpt-4")
 
-        history_aware_retriever = create_history_aware_retriever(llm,retriever, contextualize_q_prompt)
-        qa_prompt = ChatPromptTemplate.from_messages([("system", context_prompt),MessagesPlaceholder("chat_history"),("human", "{input}"),])
-        question_answer_chain =  create_stuff_documents_chain(llm, qa_prompt)
+        history_aware_retriever = create_history_aware_retriever(llm, retriever,
+                                                               contextualize_q_prompt)
+        qa_prompt = ChatPromptTemplate.from_messages([
+            ("system", context_prompt),
+            MessagesPlaceholder("chat_history"),
+            ("human", "{input}"),
+        ])
+        question_answer_chain = create_stuff_documents_chain(llm, qa_prompt)
         rag_chain = create_retrieval_chain(history_aware_retriever, question_answer_chain)
         return rag_chain
 
-    def conversation_chain(self,query: str,rag_chain,file:str =None):
+    def conversation_chain(self, query: str, rag_chain, file: str = None):
         """
-        Function to create a conversation chain
+        Create conversation chain.
+
         Args:
-            query: query to search
-            rag_chain : rag_chain model
-            file: load a file and update it with the conversation. If None it will not be saved.
+            query (str): User query
+            rag_chain: RAG chain instance
+            file (str): Optional file to save conversation
+
         Returns:
-            results
+            List: Conversation history
+
+        Example:
+            ```python
+            history = gen.conversation_chain(
+                "Tell me about...",
+                rag_chain,
+                file='conversation.txt'
+            )
+            ```
         """
         if file is not None:
             try:
-                chat_history = self.load_conversation(file,list_type=True)
+                chat_history = self.load_conversation(file, list_type=True)
                 if len(chat_history) == 0:
                     chat_history = []
             except:
                 chat_history = []
         else:
             chat_history = []
-        query = "You : " + query 
-        res = rag_chain.invoke({"input": query,"chat_history": chat_history})
+
+        query = "You : " + query
+        res = rag_chain.invoke({"input": query, "chat_history": chat_history})
         print(f"Response: {res['answer']}")
         chat_history.append(HumanMessage(content=query))
         chat_history.append(SystemMessage(content=res['answer']))
         if file is not None:
-            self.save_conversation(chat_history,file)
+            self.save_conversation(chat_history, file)
         return chat_history
 
-    def load_conversation(self,file: str,list_type: bool = False):
+    def load_conversation(self, file: str, list_type: bool = False):
         """
-        Function to load the conversation
+        Load conversation history.
+
         Args:
-            file: file to load
-            list_type: if True, return the chat_history as a list. Default is False.
+            file (str): Path to conversation file
+            list_type (bool): Whether to return as list
+
         Returns:
-            chat_history
+            Union[str, List]: Conversation history
         """
         if list_type:
             chat_history = []
-            with open(file,'r') as f:
+            with open(file, 'r') as f:
                 for line in f:
-                    # inner_list = [elt.strip() for elt in line.split(',')]
                     chat_history.append(line.strip())
         else:
             with open(file, "r") as f:
                 chat_history = f.read()
         return chat_history
 
-    def save_conversation(self,chat: str,file: str):
+    def save_conversation(self, chat: Union[str, List], file: str):
         """
-        Function to save the conversation
+        Save conversation history.
+
         Args:
-            chat: chat results
-            file: file to save
-        Returns:
-            None
+            chat (Union[str, List]): Conversation to save
+            file (str): Path to save file
         """
-        if isinstance(chat,str):
+        if isinstance(chat, str):
             with open(file, "a") as f:
                 f.write(chat)
-        elif isinstance(chat,list):
+        elif isinstance(chat, list):
             with open(file, "a") as f:
                 for i in chat[-2:]:
                     f.write("%s\n" % i)
         print(f"Saved file : {file}")
 
-    def firecrawl_web(self, website, api_key: str = None, mode="scrape", file_to_save: str = './firecrawl_embeddings',**kwargs):
+    def firecrawl_web(self, website: str, api_key: str = None, mode: str = "scrape",
+                      file_to_save: str = './firecrawl_embeddings', **kwargs):
         """
-        Function to get data from website. Use this to get data from a website and save it as embeddings/retriever. To ask questions from the website,
-          use the load_retriever and query_embeddings function.
+        Get data from website using FireCrawl.
+
         Args:
-            website : str - link to website.
-            api_key : api key of firecrawl, if None environment variable "FIRECRAWL_API_KEY" will be used.
-            mode(str) : 'scrape' default to just use the same page. Not the whole website.
-            file_to_save: path to save the embeddings
-            **kwargs: additional arguments
+            website (str): Website URL to crawl
+            api_key (str): Optional FireCrawl API key
+            mode (str): Crawl mode (default: "scrape")
+            file_to_save (str): Path to save embeddings
+            **kwargs: Additional arguments for FireCrawl
+
         Returns:
-            retriever
+            Chroma: Vector store with crawled data
+
+        Example:
+            ```python
+            db = gen.firecrawl_web(
+                "https://example.com",
+                mode="scrape",
+                file_to_save='./crawl_embeddings'
+            )
+            ```
         """
-        if not check_package("firecrawl"):
-            raise ImportError("Firecrawl package not found. Please install it using: pip install firecrawl")
-        
+        if not ModelProvider.check_package("firecrawl"):
+            raise ImportError("Firecrawl package not found. Please install: pip install firecrawl")
+
         if api_key is None:
             api_key = os.getenv("FIRECRAWL_API_KEY")
+
         loader = FireCrawlLoader(api_key=api_key, url=website, mode=mode)
         docs = loader.load()
+
         for doc in docs:
             for key, value in doc.metadata.items():
                 if isinstance(value, list):
                     doc.metadata[key] = ", ".join(map(str, value))
+
         text_splitter = CharacterTextSplitter(chunk_size=1000, chunk_overlap=0)
         split_docs = text_splitter.split_documents(docs)
+
         print("\n--- Document Chunks Information ---")
         print(f"Number of document chunks: {len(split_docs)}")
         print(f"Sample chunk:\n{split_docs[0].page_content}\n")
+
         embeddings = self.model
-        db = Chroma.from_documents(
-            split_docs, embeddings, persist_directory=file_to_save)        
+        db = Chroma.from_documents(split_docs, embeddings,
+                                 persist_directory=file_to_save)
         print(f"Retriever saved at {file_to_save}")
         return db

mb_rag/version.py

@@ -1,5 +1,5 @@
 MAJOR_VERSION = 1
 MINOR_VERSION = 0
-PATCH_VERSION = 123
+PATCH_VERSION = 125
 version = '{}.{}.{}'.format(MAJOR_VERSION, MINOR_VERSION, PATCH_VERSION)
 __all__  = ['MAJOR_VERSION', 'MINOR_VERSION', 'PATCH_VERSION', 'version']

{mb_rag-1.0.123.dist-info → mb_rag-1.0.125.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: mb_rag
-Version: 1.0.123
+Version: 1.0.125
 Summary: RAG function file
 Author: ['Malav Bateriwala']
 Requires-Python: >=3.8

{mb_rag-1.0.123.dist-info → mb_rag-1.0.125.dist-info}/RECORD

@@ -1,15 +1,15 @@
 mb_rag/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-mb_rag/version.py,sha256=ZiSwRI0jkOCO-hb51gEQ--3dbft2G13KKjEohjZMcOA,208
+mb_rag/version.py,sha256=XPsqrBre0W3TXoTeuLAD2FpvSnzAgbTcMVKLDjbbv20,208
 mb_rag/chatbot/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-mb_rag/chatbot/basic.py,sha256=vo-RrXfp2oHxT2TUGfsgwWfq-ozPKCNA51ZRpgHkd_s,14177
+mb_rag/chatbot/basic.py,sha256=OR2IvDg-Sy968C2Mna6lxmFfh7Czj8yCEkCfyvtxBwI,14223
 mb_rag/chatbot/chains.py,sha256=vDbLX5R29sWN1pcFqJ5fyxJEgMCM81JAikunAEvMC9A,7223
 mb_rag/chatbot/prompts.py,sha256=n1PyiLbU-5fkslRv6aVOzt0dDlwya_cEdQ7kRnRhMuY,1749
 mb_rag/rag/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-mb_rag/rag/embeddings.py,sha256=kOnHjrbi0GRVErfcjML8fZz-KttipObUfa5fW9tGOoY,21196
+mb_rag/rag/embeddings.py,sha256=E1USUON80Y5VzPENWHx9ke2FNpe6w-nxeddfJl9kOR8,27202
 mb_rag/utils/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 mb_rag/utils/bounding_box.py,sha256=XnuhnLrsGvsI8P8VtOwlBrDlFE2It1HEZOcLlK6kusE,7931
 mb_rag/utils/extra.py,sha256=spbFrGgdruNyYQ5PzgvpSIa6Nm0rn9bb4qc8W9g582o,2492
-mb_rag-1.0.123.dist-info/METADATA,sha256=TJxZqi8GX3o5b_BhE_67v-eFJvkwAGtoyB8kV1yByEs,154
-mb_rag-1.0.123.dist-info/WHEEL,sha256=PZUExdf71Ui_so67QXpySuHtCi3-J3wvF4ORK6k_S8U,91
-mb_rag-1.0.123.dist-info/top_level.txt,sha256=FIK1eAa5uYnurgXZquBG-s3PIy-HDTC5yJBW4lTH_pM,7
-mb_rag-1.0.123.dist-info/RECORD,,
+mb_rag-1.0.125.dist-info/METADATA,sha256=D5VYXjhHmVZE3UREC0cMov6z23X7HAJlyaGbLCpfYYY,154
+mb_rag-1.0.125.dist-info/WHEEL,sha256=PZUExdf71Ui_so67QXpySuHtCi3-J3wvF4ORK6k_S8U,91
+mb_rag-1.0.125.dist-info/top_level.txt,sha256=FIK1eAa5uYnurgXZquBG-s3PIy-HDTC5yJBW4lTH_pM,7
+mb_rag-1.0.125.dist-info/RECORD,,