import os from importlib.metadata import version from inspect import currentframe, getframeinfo from pathlib import Path from decouple import config from theflow.settings.default import * # noqa cur_frame = currentframe() if cur_frame is None: raise ValueError("Cannot get the current frame.") this_file = getframeinfo(cur_frame).filename this_dir = Path(this_file).parent # change this if your app use a different name KH_PACKAGE_NAME = "kotaemon_app" KH_APP_VERSION = config("KH_APP_VERSION", "local") if not KH_APP_VERSION: try: # Caution: This might produce the wrong version # https://stackoverflow.com/a/59533071 KH_APP_VERSION = version(KH_PACKAGE_NAME) except Exception as e: print(f"Failed to get app version: {e}") # App can be ran from anywhere and it's not trivial to decide where to store app data. # So let's use the same directory as the flowsetting.py file. # KH_APP_DATA_DIR = this_dir / "ktem_app_data" # override app data dir to fit preview data KH_APP_DATA_DIR = Path("/home/ubuntu/lib-knowledgehub/kotaemon/ktem_app_data") KH_APP_DATA_DIR.mkdir(parents=True, exist_ok=True) # User data directory KH_USER_DATA_DIR = KH_APP_DATA_DIR / "user_data" KH_USER_DATA_DIR.mkdir(parents=True, exist_ok=True) # markdowm output directory KH_MARKDOWN_OUTPUT_DIR = KH_APP_DATA_DIR / "markdown_cache_dir" KH_MARKDOWN_OUTPUT_DIR.mkdir(parents=True, exist_ok=True) # chunks output directory KH_CHUNKS_OUTPUT_DIR = KH_APP_DATA_DIR / "chunks_cache_dir" KH_CHUNKS_OUTPUT_DIR.mkdir(parents=True, exist_ok=True) # zip output directory KH_ZIP_OUTPUT_DIR = KH_APP_DATA_DIR / "zip_cache_dir" KH_ZIP_OUTPUT_DIR.mkdir(parents=True, exist_ok=True) # zip input directory KH_ZIP_INPUT_DIR = KH_APP_DATA_DIR / "zip_cache_dir_in" KH_ZIP_INPUT_DIR.mkdir(parents=True, exist_ok=True) # HF models can be big, let's store them in the app data directory so that it's easier # for users to manage their storage. # ref: https://huggingface.co/docs/huggingface_hub/en/guides/manage-cache os.environ["HF_HOME"] = str(KH_APP_DATA_DIR / "huggingface") os.environ["HF_HUB_CACHE"] = str(KH_APP_DATA_DIR / "huggingface") # doc directory KH_DOC_DIR = this_dir / "docs" KH_MODE = "dev" KH_FEATURE_USER_MANAGEMENT = False KH_USER_CAN_SEE_PUBLIC = None KH_FEATURE_USER_MANAGEMENT_ADMIN = str( config("KH_FEATURE_USER_MANAGEMENT_ADMIN", default="admin") ) KH_FEATURE_USER_MANAGEMENT_PASSWORD = str( config("KH_FEATURE_USER_MANAGEMENT_PASSWORD", default="admin") ) KH_ENABLE_ALEMBIC = False KH_DATABASE = f"sqlite:///file:{KH_USER_DATA_DIR / 'sql.db?mode=ro&uri=true'}" KH_FILESTORAGE_PATH = str(KH_USER_DATA_DIR / "files") KH_DOCSTORE = { # "__type__": "kotaemon.storages.ElasticsearchDocumentStore", # "__type__": "kotaemon.storages.SimpleFileDocumentStore", "__type__": "kotaemon.storages.LanceDBDocumentStore", "path": str(KH_USER_DATA_DIR / "docstore"), } KH_VECTORSTORE = { # "__type__": "kotaemon.storages.LanceDBVectorStore", "__type__": "kotaemon.storages.ChromaVectorStore", "path": str(KH_USER_DATA_DIR / "vectorstore"), } KH_LLMS = {} KH_EMBEDDINGS = {} # populate options from config if config("AZURE_OPENAI_API_KEY", default="") and config( "AZURE_OPENAI_ENDPOINT", default="" ): if config("AZURE_OPENAI_CHAT_DEPLOYMENT", default=""): KH_LLMS["azure"] = { "spec": { "__type__": "kotaemon.llms.AzureChatOpenAI", "temperature": 0, "azure_endpoint": config("AZURE_OPENAI_ENDPOINT", default=""), "api_key": config("AZURE_OPENAI_API_KEY", default=""), "api_version": config("OPENAI_API_VERSION", default="") or "2024-02-15-preview", "azure_deployment": config("AZURE_OPENAI_CHAT_DEPLOYMENT", default=""), "timeout": 20, }, "default": False, } if config("AZURE_OPENAI_EMBEDDINGS_DEPLOYMENT", default=""): KH_EMBEDDINGS["azure"] = { "spec": { "__type__": "kotaemon.embeddings.AzureOpenAIEmbeddings", "azure_endpoint": config("AZURE_OPENAI_ENDPOINT", default=""), "api_key": config("AZURE_OPENAI_API_KEY", default=""), "api_version": config("OPENAI_API_VERSION", default="") or "2024-02-15-preview", "azure_deployment": config( "AZURE_OPENAI_EMBEDDINGS_DEPLOYMENT", default="" ), "timeout": 10, }, "default": False, } if config("OPENAI_API_KEY", default=""): KH_LLMS["openai"] = { "spec": { "__type__": "kotaemon.llms.ChatOpenAI", "temperature": 0, "base_url": config("OPENAI_API_BASE", default="") or "https://api.openai.com/v1", "api_key": config("OPENAI_API_KEY", default=""), "model": config("OPENAI_CHAT_MODEL", default="gpt-3.5-turbo"), "timeout": 20, }, "default": True, } KH_EMBEDDINGS["openai"] = { "spec": { "__type__": "kotaemon.embeddings.OpenAIEmbeddings", "base_url": config("OPENAI_API_BASE", default="https://api.openai.com/v1"), "api_key": config("OPENAI_API_KEY", default=""), "model": config( "OPENAI_EMBEDDINGS_MODEL", default="text-embedding-ada-002" ), "timeout": 10, "context_length": 8191, }, "default": True, } if config("LOCAL_MODEL", default=""): KH_LLMS["ollama"] = { "spec": { "__type__": "kotaemon.llms.ChatOpenAI", "base_url": "http://localhost:11434/v1/", "model": config("LOCAL_MODEL", default="llama3.1:8b"), }, "default": False, } KH_EMBEDDINGS["ollama"] = { "spec": { "__type__": "kotaemon.embeddings.OpenAIEmbeddings", "base_url": "http://localhost:11434/v1/", "model": config("LOCAL_MODEL_EMBEDDINGS", default="nomic-embed-text"), }, "default": False, } KH_EMBEDDINGS["local-bge-en"] = { "spec": { "__type__": "kotaemon.embeddings.FastEmbedEmbeddings", "model_name": "BAAI/bge-base-en-v1.5", }, "default": False, } KH_REASONINGS = [ "ktem.reasoning.simple.FullQAPipeline", "ktem.reasoning.simple.FullDecomposeQAPipeline", "ktem.reasoning.react.ReactAgentPipeline", "ktem.reasoning.rewoo.RewooAgentPipeline", ] KH_REASONINGS_USE_MULTIMODAL = False KH_VLM_ENDPOINT = "{0}/openai/deployments/{1}/chat/completions?api-version={2}".format( config("AZURE_OPENAI_ENDPOINT", default=""), config("OPENAI_VISION_DEPLOYMENT_NAME", default="gpt-4o"), config("OPENAI_API_VERSION", default=""), ) SETTINGS_APP: dict[str, dict] = {} SETTINGS_REASONING = { "use": { "name": "Reasoning options", "value": None, "choices": [], "component": "radio", }, "lang": { "name": "Language", "value": "en", "choices": [("English", "en"), ("Japanese", "ja"), ("Vietnamese", "vi")], "component": "dropdown", }, "max_context_length": { "name": "Max context length (LLM)", "value": 32000, "component": "number", }, } KH_INDEX_TYPES = [ "ktem.index.file.FileIndex", "ktem.index.file.graph.GraphRAGIndex", ] KH_INDICES = [ { "name": "File", "config": { "supported_file_types": ( ".png, .jpeg, .jpg, .tiff, .tif, .pdf, .xls, .xlsx, .doc, .docx, " ".pptx, .csv, .html, .mhtml, .txt, .zip" ), "private": False, }, "index_type": "ktem.index.file.FileIndex", }, { "name": "GraphRAG", "config": { "supported_file_types": ( ".png, .jpeg, .jpg, .tiff, .tif, .pdf, .xls, .xlsx, .doc, .docx, " ".pptx, .csv, .html, .mhtml, .txt, .zip" ), "private": False, }, "index_type": "ktem.index.file.graph.GraphRAGIndex", }, ]