Skip to main content

​
Class: LangChatConfig

Configuration class for LangChat. All settings can be customized.

​
Constructor

LangChatConfig(
    openai_api_keys: List[str],
    openai_model: str = "gpt-4o-mini",
    openai_temperature: float = 1.0,
    openai_embedding_model: str = "text-embedding-3-large",
    pinecone_api_key: Optional[str] = None,
    pinecone_index_name: Optional[str] = None,
    supabase_url: Optional[str] = None,
    supabase_key: Optional[str] = None,
    retrieval_k: int = 5,
    reranker_top_n: int = 3,
    reranker_model: str = "ms-marco-MiniLM-L-12-v2",
    reranker_cache_dir: str = "rerank_models",
    max_chat_history: int = 20,
    memory_window: int = 20,
    timezone: str = "Asia/Dhaka",
    system_prompt_template: Optional[str] = None,
    standalone_question_prompt: Optional[str] = None,
    max_llm_retries: int = 2,
    server_port: int = 8000
)
Parameters:
​
openai_api_keys
List[str]
required
List of OpenAI API keys. Can provide multiple keys for automatic rotation.
​
openai_model
str
default:"gpt-4o-mini"
OpenAI model to use. Options: “gpt-4o-mini”, “gpt-4o”, “gpt-4-turbo”, “gpt-3.5-turbo”
​
openai_temperature
float
default:"1.0"
Model temperature (0.0-2.0). Higher = more creative, lower = more focused
​
openai_embedding_model
str
default:"text-embedding-3-large"
OpenAI embedding model. Options: “text-embedding-3-large”, “text-embedding-3-small”, “text-embedding-ada-002”
​
pinecone_api_key
str | None
default:"None"
Pinecone API key
​
pinecone_index_name
str | None
default:"None"
Pinecone index name (must be pre-created)
​
supabase_url
str | None
default:"None"
Supabase project URL
​
supabase_key
str | None
default:"None"
Supabase API key (anon or service role)
​
retrieval_k
int
default:"5"
Number of documents to retrieve from vector search
​
reranker_top_n
int
default:"3"
Number of top documents after reranking
​
reranker_model
str
default:"ms-marco-MiniLM-L-12-v2"
Flashrank reranker model name
​
reranker_cache_dir
str
default:"rerank_models"
Directory to cache reranker models
​
max_chat_history
int
default:"20"
Maximum chat history messages to keep in memory
​
memory_window
int
default:"20"
Conversation buffer window size
​
timezone
str
default:"Asia/Dhaka"
Timezone for date/time formatting
​
system_prompt_template
str | None
default:"None"
Custom system prompt template. Uses default if None.
​
standalone_question_prompt
str | None
default:"None"
Custom standalone question prompt. Uses default if None.
​
max_llm_retries
int
default:"2"
Maximum retry attempts per API key
​
server_port
int
default:"8000"
Port for API server

​
Class Methods

​
from_env()

Create configuration from environment variables. 
@classmethod
def from_env() -> LangChatConfig
Returns: LangChatConfig - Configuration loaded from environment variables Example: 
from langchat import LangChatConfig

# Load from environment variables
config = LangChatConfig.from_env()
Required Environment Variables:
  • OPENAI_API_KEYS or OPENAI_API_KEY - OpenAI API key(s), comma-separated
  • PINECONE_API_KEY - Pinecone API key
  • PINECONE_INDEX_NAME - Pinecone index name
  • SUPABASE_URL - Supabase project URL
  • SUPABASE_KEY - Supabase API key
Optional Environment Variables:
  • OPENAI_MODEL - OpenAI model (default: “gpt-4o-mini”)
  • OPENAI_TEMPERATURE - Model temperature (default: “1.0”)
  • OPENAI_EMBEDDING_MODEL - Embedding model (default: “text-embedding-3-large”)
  • RETRIEVAL_K - Retrieval count (default: “5”)
  • RERANKER_TOP_N - Reranker top N (default: “3”)
  • MAX_CHAT_HISTORY - Max history (default: “20”)
  • SERVER_PORT or PORT - Server port (default: “8000”)
  • TIMEZONE - Timezone (default: “Asia/Dhaka”)

​
Instance Methods

​
get_formatted_time()

Get current formatted time based on configured timezone. 
def get_formatted_time(self) -> str
Returns: str - Formatted time string (e.g., “Monday, 01 January 2024”) Example: 
config = LangChatConfig.from_env()
time_str = config.get_formatted_time()
print(time_str)  # "Monday, 01 January 2024"

​
get_default_prompt_template()

Get default system prompt template. 
def get_default_prompt_template(self) -> str
Returns: str - Default prompt template string Example: 
config = LangChatConfig.from_env()
default_prompt = config.get_default_prompt_template()
print(default_prompt)

​
Usage Examples

​
Basic Configuration

from langchat import LangChatConfig

config = LangChatConfig(
    openai_api_keys=["sk-..."],
    pinecone_api_key="pcsk-...",
    pinecone_index_path="my-index",
    supabase_url="https://xxxxx.supabase.co",
    supabase_key="eyJ..."
)

​
From Environment Variables

from langchat import LangChatConfig

# Set environment variables first
import os
os.environ["OPENAI_API_KEYS"] = "sk-key1,sk-key2"
os.environ["PINECONE_API_KEY"] = "pcsk-..."
os.environ["PINECONE_INDEX_NAME"] = "my-index"
os.environ["SUPABASE_URL"] = "https://xxxxx.supabase.co"
os.environ["SUPABASE_KEY"] = "eyJ..."

# Load configuration
config = LangChatConfig.from_env()

​
Production Configuration

import os
from langchat import LangChatConfig

config = LangChatConfig(
    # Multiple API keys for rotation
    openai_api_keys=os.getenv("OPENAI_API_KEYS").split(","),
    
    # Model configuration
    openai_model="gpt-4o-mini",
    openai_temperature=0.8,
    openai_embedding_model="text-embedding-3-large",
    max_llm_retries=2,
    
    # Vector search
    pinecone_api_key=os.getenv("PINECONE_API_KEY"),
    pinecone_index_name=os.getenv("PINECONE_INDEX_NAME"),
    retrieval_k=10,
    reranker_top_n=5,
    
    # Database
    supabase_url=os.getenv("SUPABASE_URL"),
    supabase_key=os.getenv("SUPABASE_KEY"),
    
    # Session
    max_chat_history=50,
    memory_window=50,
    
    # Server
    server_port=int(os.getenv("PORT", "8000"))
)

​
Custom Prompts

from langchat import LangChatConfig

config = LangChatConfig(
    # ... other config ...
    
    # Custom system prompt
    system_prompt_template="""You are a helpful assistant.
    
    Use the following context:
    {context}
    
    Chat history: {chat_history}
    Question: {question}
    Answer:""",
    
    # Custom standalone question prompt
    standalone_question_prompt="""Convert this to a standalone query:
    
    Chat History: {chat_history}
    Question: {question}
    Standalone query:"""
)

​
Hybrid Configuration

from langchat import LangChatConfig

# Load from environment
config = LangChatConfig.from_env()

# Override specific settings
config.openai_model = "gpt-4"
config.server_port = 8080
config.retrieval_k = 15

​
Validation

Configuration is validated when LangChat is initialized: 
from langchat import LangChat, LangChatConfig

try:
    config = LangChatConfig(
        openai_api_keys=[],  # Empty list
        # ... other config ...
    )
    langchat = LangChat(config=config)
except ValueError as e:
    print(f"Configuration error: {e}")
Common Validation Errors:
  • "OpenAI API keys must be provided" - Empty or missing API keys
  • "Supabase URL and key must be provided" - Missing Supabase credentials
  • "Pinecone API key must be provided" - Missing Pinecone key
  • "Pinecone index name must be provided" - Missing index name