Merge pull request #2376 from open-webui/dev

0.1.125

backend/config.py

@@ -5,6 +5,7 @@ import chromadb
 from chromadb import Settings
 from base64 import b64encode
 from bs4 import BeautifulSoup
+from typing import TypeVar, Generic, Union
 
 from pathlib import Path
 import json
@@ -17,7 +18,6 @@ import shutil
 from secrets import token_bytes
 from constants import ERROR_MESSAGES
 
-
 ####################################
 # Load .env file
 ####################################
@@ -71,7 +71,6 @@ for source in log_sources:
 
 log.setLevel(SRC_LOG_LEVELS["CONFIG"])
 
-
 WEBUI_NAME = os.environ.get("WEBUI_NAME", "Open WebUI")
 if WEBUI_NAME != "Open WebUI":
     WEBUI_NAME += " (Open WebUI)"
@@ -161,16 +160,6 @@ CHANGELOG = changelog_json
 
 WEBUI_VERSION = os.environ.get("WEBUI_VERSION", "v1.0.0-alpha.100")
 
-####################################
-# WEBUI_AUTH (Required for security)
-####################################
-
-WEBUI_AUTH = os.environ.get("WEBUI_AUTH", "True").lower() == "true"
-WEBUI_AUTH_TRUSTED_EMAIL_HEADER = os.environ.get(
-    "WEBUI_AUTH_TRUSTED_EMAIL_HEADER", None
-)
-
-
 ####################################
 # DATA/FRONTEND BUILD DIR
 ####################################
@@ -184,6 +173,108 @@ try:
 except:
     CONFIG_DATA = {}
 
+
+####################################
+# Config helpers
+####################################
+
+
+def save_config():
+    try:
+        with open(f"{DATA_DIR}/config.json", "w") as f:
+            json.dump(CONFIG_DATA, f, indent="\t")
+    except Exception as e:
+        log.exception(e)
+
+
+def get_config_value(config_path: str):
+    path_parts = config_path.split(".")
+    cur_config = CONFIG_DATA
+    for key in path_parts:
+        if key in cur_config:
+            cur_config = cur_config[key]
+        else:
+            return None
+    return cur_config
+
+
+T = TypeVar("T")
+
+
+class PersistentConfig(Generic[T]):
+    def __init__(self, env_name: str, config_path: str, env_value: T):
+        self.env_name = env_name
+        self.config_path = config_path
+        self.env_value = env_value
+        self.config_value = get_config_value(config_path)
+        if self.config_value is not None:
+            log.info(f"'{env_name}' loaded from config.json")
+            self.value = self.config_value
+        else:
+            self.value = env_value
+
+    def __str__(self):
+        return str(self.value)
+
+    @property
+    def __dict__(self):
+        raise TypeError(
+            "PersistentConfig object cannot be converted to dict, use config_get or .value instead."
+        )
+
+    def __getattribute__(self, item):
+        if item == "__dict__":
+            raise TypeError(
+                "PersistentConfig object cannot be converted to dict, use config_get or .value instead."
+            )
+        return super().__getattribute__(item)
+
+    def save(self):
+        # Don't save if the value is the same as the env value and the config value
+        if self.env_value == self.value:
+            if self.config_value == self.value:
+                return
+        log.info(f"Saving '{self.env_name}' to config.json")
+        path_parts = self.config_path.split(".")
+        config = CONFIG_DATA
+        for key in path_parts[:-1]:
+            if key not in config:
+                config[key] = {}
+            config = config[key]
+        config[path_parts[-1]] = self.value
+        save_config()
+        self.config_value = self.value
+
+
+class AppConfig:
+    _state: dict[str, PersistentConfig]
+
+    def __init__(self):
+        super().__setattr__("_state", {})
+
+    def __setattr__(self, key, value):
+        if isinstance(value, PersistentConfig):
+            self._state[key] = value
+        else:
+            self._state[key].value = value
+            self._state[key].save()
+
+    def __getattr__(self, key):
+        return self._state[key].value
+
+
+####################################
+# WEBUI_AUTH (Required for security)
+####################################
+
+WEBUI_AUTH = os.environ.get("WEBUI_AUTH", "True").lower() == "true"
+WEBUI_AUTH_TRUSTED_EMAIL_HEADER = os.environ.get(
+    "WEBUI_AUTH_TRUSTED_EMAIL_HEADER", None
+)
+JWT_EXPIRES_IN = PersistentConfig(
+    "JWT_EXPIRES_IN", "auth.jwt_expiry", os.environ.get("JWT_EXPIRES_IN", "-1")
+)
+
 ####################################
 # Static DIR
 ####################################
@@ -318,12 +409,22 @@ OLLAMA_BASE_URLS = os.environ.get("OLLAMA_BASE_URLS", "")
 OLLAMA_BASE_URLS = OLLAMA_BASE_URLS if OLLAMA_BASE_URLS != "" else OLLAMA_BASE_URL
 
 OLLAMA_BASE_URLS = [url.strip() for url in OLLAMA_BASE_URLS.split(";")]
-
+OLLAMA_BASE_URLS = PersistentConfig(
+    "OLLAMA_BASE_URLS", "ollama.base_urls", OLLAMA_BASE_URLS
+)
 
 ####################################
 # OPENAI_API
 ####################################
 
+
+ENABLE_OPENAI_API = PersistentConfig(
+    "ENABLE_OPENAI_API",
+    "openai.enable",
+    os.environ.get("ENABLE_OPENAI_API", "True").lower() == "true",
+)
+
+
 OPENAI_API_KEY = os.environ.get("OPENAI_API_KEY", "")
 OPENAI_API_BASE_URL = os.environ.get("OPENAI_API_BASE_URL", "")
 
@@ -335,7 +436,9 @@ OPENAI_API_KEYS = os.environ.get("OPENAI_API_KEYS", "")
 OPENAI_API_KEYS = OPENAI_API_KEYS if OPENAI_API_KEYS != "" else OPENAI_API_KEY
 
 OPENAI_API_KEYS = [url.strip() for url in OPENAI_API_KEYS.split(";")]
-
+OPENAI_API_KEYS = PersistentConfig(
+    "OPENAI_API_KEYS", "openai.api_keys", OPENAI_API_KEYS
+)
 
 OPENAI_API_BASE_URLS = os.environ.get("OPENAI_API_BASE_URLS", "")
 OPENAI_API_BASE_URLS = (
@@ -346,37 +449,42 @@ OPENAI_API_BASE_URLS = [
     url.strip() if url != "" else "https://api.openai.com/v1"
     for url in OPENAI_API_BASE_URLS.split(";")
 ]
+OPENAI_API_BASE_URLS = PersistentConfig(
+    "OPENAI_API_BASE_URLS", "openai.api_base_urls", OPENAI_API_BASE_URLS
+)
 
 OPENAI_API_KEY = ""
 
 try:
-    OPENAI_API_KEY = OPENAI_API_KEYS[
-        OPENAI_API_BASE_URLS.index("https://api.openai.com/v1")
+    OPENAI_API_KEY = OPENAI_API_KEYS.value[
+        OPENAI_API_BASE_URLS.value.index("https://api.openai.com/v1")
     ]
 except:
     pass
 
 OPENAI_API_BASE_URL = "https://api.openai.com/v1"
 
-
 ####################################
 # WEBUI
 ####################################
 
-ENABLE_SIGNUP = (
+ENABLE_SIGNUP = PersistentConfig(
+    "ENABLE_SIGNUP",
+    "ui.enable_signup",
+    (
         False
-    if WEBUI_AUTH == False
+        if not WEBUI_AUTH
         else os.environ.get("ENABLE_SIGNUP", "True").lower() == "true"
+    ),
+)
+DEFAULT_MODELS = PersistentConfig(
+    "DEFAULT_MODELS", "ui.default_models", os.environ.get("DEFAULT_MODELS", None)
 )
-DEFAULT_MODELS = os.environ.get("DEFAULT_MODELS", None)
-
 
-DEFAULT_PROMPT_SUGGESTIONS = (
-    CONFIG_DATA["ui"]["prompt_suggestions"]
-    if "ui" in CONFIG_DATA
-    and "prompt_suggestions" in CONFIG_DATA["ui"]
-    and type(CONFIG_DATA["ui"]["prompt_suggestions"]) is list
-    else [
+DEFAULT_PROMPT_SUGGESTIONS = PersistentConfig(
+    "DEFAULT_PROMPT_SUGGESTIONS",
+    "ui.prompt_suggestions",
+    [
         {
             "title": ["Help me study", "vocabulary for a college entrance exam"],
             "content": "Help me study vocabulary: write a sentence for me to fill in the blank, and I'll try to pick the correct option.",
@@ -404,23 +512,40 @@ DEFAULT_PROMPT_SUGGESTIONS = (
             "title": ["Overcome procrastination", "give me tips"],
             "content": "Could you start by asking me about instances when I procrastinate the most and then give me some suggestions to overcome it?",
         },
-    ]
+    ],
 )
 
-
-DEFAULT_USER_ROLE = os.getenv("DEFAULT_USER_ROLE", "pending")
+DEFAULT_USER_ROLE = PersistentConfig(
+    "DEFAULT_USER_ROLE",
+    "ui.default_user_role",
+    os.getenv("DEFAULT_USER_ROLE", "pending"),
+)
 
 USER_PERMISSIONS_CHAT_DELETION = (
     os.environ.get("USER_PERMISSIONS_CHAT_DELETION", "True").lower() == "true"
 )
 
-USER_PERMISSIONS = {"chat": {"deletion": USER_PERMISSIONS_CHAT_DELETION}}
+USER_PERMISSIONS = PersistentConfig(
+    "USER_PERMISSIONS",
+    "ui.user_permissions",
+    {"chat": {"deletion": USER_PERMISSIONS_CHAT_DELETION}},
+)
 
-ENABLE_MODEL_FILTER = os.environ.get("ENABLE_MODEL_FILTER", "False").lower() == "true"
+ENABLE_MODEL_FILTER = PersistentConfig(
+    "ENABLE_MODEL_FILTER",
+    "model_filter.enable",
+    os.environ.get("ENABLE_MODEL_FILTER", "False").lower() == "true",
+)
 MODEL_FILTER_LIST = os.environ.get("MODEL_FILTER_LIST", "")
-MODEL_FILTER_LIST = [model.strip() for model in MODEL_FILTER_LIST.split(";")]
+MODEL_FILTER_LIST = PersistentConfig(
+    "MODEL_FILTER_LIST",
+    "model_filter.list",
+    [model.strip() for model in MODEL_FILTER_LIST.split(";")],
+)
 
-WEBHOOK_URL = os.environ.get("WEBHOOK_URL", "")
+WEBHOOK_URL = PersistentConfig(
+    "WEBHOOK_URL", "webhook_url", os.environ.get("WEBHOOK_URL", "")
+)
 
 ENABLE_ADMIN_EXPORT = os.environ.get("ENABLE_ADMIN_EXPORT", "True").lower() == "true"
 
@@ -458,26 +583,45 @@ else:
 CHROMA_HTTP_SSL = os.environ.get("CHROMA_HTTP_SSL", "false").lower() == "true"
 # this uses the model defined in the Dockerfile ENV variable. If you dont use docker or docker based deployments such as k8s, the default embedding model will be used (sentence-transformers/all-MiniLM-L6-v2)
 
-RAG_TOP_K = int(os.environ.get("RAG_TOP_K", "5"))
-RAG_RELEVANCE_THRESHOLD = float(os.environ.get("RAG_RELEVANCE_THRESHOLD", "0.0"))
-
-ENABLE_RAG_HYBRID_SEARCH = (
-    os.environ.get("ENABLE_RAG_HYBRID_SEARCH", "").lower() == "true"
+RAG_TOP_K = PersistentConfig(
+    "RAG_TOP_K", "rag.top_k", int(os.environ.get("RAG_TOP_K", "5"))
+)
+RAG_RELEVANCE_THRESHOLD = PersistentConfig(
+    "RAG_RELEVANCE_THRESHOLD",
+    "rag.relevance_threshold",
+    float(os.environ.get("RAG_RELEVANCE_THRESHOLD", "0.0")),
 )
 
+ENABLE_RAG_HYBRID_SEARCH = PersistentConfig(
+    "ENABLE_RAG_HYBRID_SEARCH",
+    "rag.enable_hybrid_search",
+    os.environ.get("ENABLE_RAG_HYBRID_SEARCH", "").lower() == "true",
+)
 
-ENABLE_RAG_WEB_LOADER_SSL_VERIFICATION = (
-    os.environ.get("ENABLE_RAG_WEB_LOADER_SSL_VERIFICATION", "True").lower() == "true"
+ENABLE_RAG_WEB_LOADER_SSL_VERIFICATION = PersistentConfig(
+    "ENABLE_RAG_WEB_LOADER_SSL_VERIFICATION",
+    "rag.enable_web_loader_ssl_verification",
+    os.environ.get("ENABLE_RAG_WEB_LOADER_SSL_VERIFICATION", "True").lower() == "true",
 )
 
-RAG_EMBEDDING_ENGINE = os.environ.get("RAG_EMBEDDING_ENGINE", "")
+RAG_EMBEDDING_ENGINE = PersistentConfig(
+    "RAG_EMBEDDING_ENGINE",
+    "rag.embedding_engine",
+    os.environ.get("RAG_EMBEDDING_ENGINE", ""),
+)
 
-PDF_EXTRACT_IMAGES = os.environ.get("PDF_EXTRACT_IMAGES", "False").lower() == "true"
+PDF_EXTRACT_IMAGES = PersistentConfig(
+    "PDF_EXTRACT_IMAGES",
+    "rag.pdf_extract_images",
+    os.environ.get("PDF_EXTRACT_IMAGES", "False").lower() == "true",
+)
 
-RAG_EMBEDDING_MODEL = os.environ.get(
-    "RAG_EMBEDDING_MODEL", "sentence-transformers/all-MiniLM-L6-v2"
+RAG_EMBEDDING_MODEL = PersistentConfig(
+    "RAG_EMBEDDING_MODEL",
+    "rag.embedding_model",
+    os.environ.get("RAG_EMBEDDING_MODEL", "sentence-transformers/all-MiniLM-L6-v2"),
 )
-log.info(f"Embedding model set: {RAG_EMBEDDING_MODEL}"),
+log.info(f"Embedding model set: {RAG_EMBEDDING_MODEL.value}"),
 
 RAG_EMBEDDING_MODEL_AUTO_UPDATE = (
     os.environ.get("RAG_EMBEDDING_MODEL_AUTO_UPDATE", "").lower() == "true"
@@ -487,9 +631,13 @@ RAG_EMBEDDING_MODEL_TRUST_REMOTE_CODE = (
     os.environ.get("RAG_EMBEDDING_MODEL_TRUST_REMOTE_CODE", "").lower() == "true"
 )
 
-RAG_RERANKING_MODEL = os.environ.get("RAG_RERANKING_MODEL", "")
-if not RAG_RERANKING_MODEL == "":
-    log.info(f"Reranking model set: {RAG_RERANKING_MODEL}"),
+RAG_RERANKING_MODEL = PersistentConfig(
+    "RAG_RERANKING_MODEL",
+    "rag.reranking_model",
+    os.environ.get("RAG_RERANKING_MODEL", ""),
+)
+if RAG_RERANKING_MODEL.value != "":
+    log.info(f"Reranking model set: {RAG_RERANKING_MODEL.value}"),
 
 RAG_RERANKING_MODEL_AUTO_UPDATE = (
     os.environ.get("RAG_RERANKING_MODEL_AUTO_UPDATE", "").lower() == "true"
@@ -527,9 +675,14 @@ if USE_CUDA.lower() == "true":
 else:
     DEVICE_TYPE = "cpu"
 
-
-CHUNK_SIZE = int(os.environ.get("CHUNK_SIZE", "1500"))
-CHUNK_OVERLAP = int(os.environ.get("CHUNK_OVERLAP", "100"))
+CHUNK_SIZE = PersistentConfig(
+    "CHUNK_SIZE", "rag.chunk_size", int(os.environ.get("CHUNK_SIZE", "1500"))
+)
+CHUNK_OVERLAP = PersistentConfig(
+    "CHUNK_OVERLAP",
+    "rag.chunk_overlap",
+    int(os.environ.get("CHUNK_OVERLAP", "100")),
+)
 
 DEFAULT_RAG_TEMPLATE = """Use the following context as your learned knowledge, inside <context></context> XML tags.
 <context>
@@ -545,16 +698,32 @@ And answer according to the language of the user's question.
 Given the context information, answer the query.
 Query: [query]"""
 
-RAG_TEMPLATE = os.environ.get("RAG_TEMPLATE", DEFAULT_RAG_TEMPLATE)
+RAG_TEMPLATE = PersistentConfig(
+    "RAG_TEMPLATE",
+    "rag.template",
+    os.environ.get("RAG_TEMPLATE", DEFAULT_RAG_TEMPLATE),
+)
 
-RAG_OPENAI_API_BASE_URL = os.getenv("RAG_OPENAI_API_BASE_URL", OPENAI_API_BASE_URL)
-RAG_OPENAI_API_KEY = os.getenv("RAG_OPENAI_API_KEY", OPENAI_API_KEY)
+RAG_OPENAI_API_BASE_URL = PersistentConfig(
+    "RAG_OPENAI_API_BASE_URL",
+    "rag.openai_api_base_url",
+    os.getenv("RAG_OPENAI_API_BASE_URL", OPENAI_API_BASE_URL),
+)
+RAG_OPENAI_API_KEY = PersistentConfig(
+    "RAG_OPENAI_API_KEY",
+    "rag.openai_api_key",
+    os.getenv("RAG_OPENAI_API_KEY", OPENAI_API_KEY),
+)
 
 ENABLE_RAG_LOCAL_WEB_FETCH = (
     os.getenv("ENABLE_RAG_LOCAL_WEB_FETCH", "False").lower() == "true"
 )
 
-YOUTUBE_LOADER_LANGUAGE = os.getenv("YOUTUBE_LOADER_LANGUAGE", "en").split(",")
+YOUTUBE_LOADER_LANGUAGE = PersistentConfig(
+    "YOUTUBE_LOADER_LANGUAGE",
+    "rag.youtube_loader_language",
+    os.getenv("YOUTUBE_LOADER_LANGUAGE", "en").split(","),
+)
 
 ####################################
 # Transcribe
@@ -571,34 +740,78 @@ WHISPER_MODEL_AUTO_UPDATE = (
 # Images
 ####################################
 
-IMAGE_GENERATION_ENGINE = os.getenv("IMAGE_GENERATION_ENGINE", "")
+IMAGE_GENERATION_ENGINE = PersistentConfig(
+    "IMAGE_GENERATION_ENGINE",
+    "image_generation.engine",
+    os.getenv("IMAGE_GENERATION_ENGINE", ""),
+)
 
-ENABLE_IMAGE_GENERATION = (
-    os.environ.get("ENABLE_IMAGE_GENERATION", "").lower() == "true"
+ENABLE_IMAGE_GENERATION = PersistentConfig(
+    "ENABLE_IMAGE_GENERATION",
+    "image_generation.enable",
+    os.environ.get("ENABLE_IMAGE_GENERATION", "").lower() == "true",
+)
+AUTOMATIC1111_BASE_URL = PersistentConfig(
+    "AUTOMATIC1111_BASE_URL",
+    "image_generation.automatic1111.base_url",
+    os.getenv("AUTOMATIC1111_BASE_URL", ""),
 )
-AUTOMATIC1111_BASE_URL = os.getenv("AUTOMATIC1111_BASE_URL", "")
 
-COMFYUI_BASE_URL = os.getenv("COMFYUI_BASE_URL", "")
+COMFYUI_BASE_URL = PersistentConfig(
+    "COMFYUI_BASE_URL",
+    "image_generation.comfyui.base_url",
+    os.getenv("COMFYUI_BASE_URL", ""),
+)
 
-IMAGES_OPENAI_API_BASE_URL = os.getenv(
-    "IMAGES_OPENAI_API_BASE_URL", OPENAI_API_BASE_URL
+IMAGES_OPENAI_API_BASE_URL = PersistentConfig(
+    "IMAGES_OPENAI_API_BASE_URL",
+    "image_generation.openai.api_base_url",
+    os.getenv("IMAGES_OPENAI_API_BASE_URL", OPENAI_API_BASE_URL),
+)
+IMAGES_OPENAI_API_KEY = PersistentConfig(
+    "IMAGES_OPENAI_API_KEY",
+    "image_generation.openai.api_key",
+    os.getenv("IMAGES_OPENAI_API_KEY", OPENAI_API_KEY),
 )
-IMAGES_OPENAI_API_KEY = os.getenv("IMAGES_OPENAI_API_KEY", OPENAI_API_KEY)
 
-IMAGE_SIZE = os.getenv("IMAGE_SIZE", "512x512")
+IMAGE_SIZE = PersistentConfig(
+    "IMAGE_SIZE", "image_generation.size", os.getenv("IMAGE_SIZE", "512x512")
+)
 
-IMAGE_STEPS = int(os.getenv("IMAGE_STEPS", 50))
+IMAGE_STEPS = PersistentConfig(
+    "IMAGE_STEPS", "image_generation.steps", int(os.getenv("IMAGE_STEPS", 50))
+)
 
-IMAGE_GENERATION_MODEL = os.getenv("IMAGE_GENERATION_MODEL", "")
+IMAGE_GENERATION_MODEL = PersistentConfig(
+    "IMAGE_GENERATION_MODEL",
+    "image_generation.model",
+    os.getenv("IMAGE_GENERATION_MODEL", ""),
+)
 
 ####################################
 # Audio
 ####################################
 
-AUDIO_OPENAI_API_BASE_URL = os.getenv("AUDIO_OPENAI_API_BASE_URL", OPENAI_API_BASE_URL)
-AUDIO_OPENAI_API_KEY = os.getenv("AUDIO_OPENAI_API_KEY", OPENAI_API_KEY)
-AUDIO_OPENAI_API_MODEL = os.getenv("AUDIO_OPENAI_API_MODEL", "tts-1")
-AUDIO_OPENAI_API_VOICE = os.getenv("AUDIO_OPENAI_API_VOICE", "alloy")
+AUDIO_OPENAI_API_BASE_URL = PersistentConfig(
+    "AUDIO_OPENAI_API_BASE_URL",
+    "audio.openai.api_base_url",
+    os.getenv("AUDIO_OPENAI_API_BASE_URL", OPENAI_API_BASE_URL),
+)
+AUDIO_OPENAI_API_KEY = PersistentConfig(
+    "AUDIO_OPENAI_API_KEY",
+    "audio.openai.api_key",
+    os.getenv("AUDIO_OPENAI_API_KEY", OPENAI_API_KEY),
+)
+AUDIO_OPENAI_API_MODEL = PersistentConfig(
+    "AUDIO_OPENAI_API_MODEL",
+    "audio.openai.api_model",
+    os.getenv("AUDIO_OPENAI_API_MODEL", "tts-1"),
+)
+AUDIO_OPENAI_API_VOICE = PersistentConfig(
+    "AUDIO_OPENAI_API_VOICE",
+    "audio.openai.api_voice",
+    os.getenv("AUDIO_OPENAI_API_VOICE", "alloy"),
+)
 
 ####################################
 # LiteLLM

backend/main.py

+from contextlib import asynccontextmanager
 from bs4 import BeautifulSoup
 import json
 import markdown
@@ -58,6 +59,7 @@ from config import (
     SRC_LOG_LEVELS,
     WEBHOOK_URL,
     ENABLE_ADMIN_EXPORT,
+    AppConfig,
 )
 from constants import ERROR_MESSAGES
 
@@ -92,16 +94,41 @@ https://github.com/open-webui/open-webui
 """
 )
 
-app = FastAPI(docs_url="/docs" if ENV == "dev" else None, redoc_url=None)
 
-app.state.ENABLE_MODEL_FILTER = ENABLE_MODEL_FILTER
-app.state.MODEL_FILTER_LIST = MODEL_FILTER_LIST
+@asynccontextmanager
+async def lifespan(app: FastAPI):
+    if ENABLE_LITELLM:
+        asyncio.create_task(start_litellm_background())
+    yield
+    if ENABLE_LITELLM:
+        await shutdown_litellm_background()
+
+
+app = FastAPI(
+    docs_url="/docs" if ENV == "dev" else None, redoc_url=None, lifespan=lifespan
+)
 
-app.state.WEBHOOK_URL = WEBHOOK_URL
+app.state.config = AppConfig()
+app.state.config.ENABLE_MODEL_FILTER = ENABLE_MODEL_FILTER
+app.state.config.MODEL_FILTER_LIST = MODEL_FILTER_LIST
+
+app.state.config.WEBHOOK_URL = WEBHOOK_URL
 
 origins = ["*"]
 
 
+# Custom middleware to add security headers
+# class SecurityHeadersMiddleware(BaseHTTPMiddleware):
+#     async def dispatch(self, request: Request, call_next):
+#         response: Response = await call_next(request)
+#         response.headers["Cross-Origin-Opener-Policy"] = "same-origin"
+#         response.headers["Cross-Origin-Embedder-Policy"] = "require-corp"
+#         return response
+
+
+# app.add_middleware(SecurityHeadersMiddleware)
+
+
 class RAGMiddleware(BaseHTTPMiddleware):
     async def dispatch(self, request: Request, call_next):
         return_citations = False
@@ -129,12 +156,12 @@ class RAGMiddleware(BaseHTTPMiddleware):
                 data["messages"], citations = rag_messages(
                     docs=data["docs"],
                     messages=data["messages"],
-                    template=rag_app.state.RAG_TEMPLATE,
+                    template=rag_app.state.config.RAG_TEMPLATE,
                     embedding_function=rag_app.state.EMBEDDING_FUNCTION,
-                    k=rag_app.state.TOP_K,
+                    k=rag_app.state.config.TOP_K,
                     reranking_function=rag_app.state.sentence_transformer_rf,
-                    r=rag_app.state.RELEVANCE_THRESHOLD,
-                    hybrid_search=rag_app.state.ENABLE_RAG_HYBRID_SEARCH,
+                    r=rag_app.state.config.RELEVANCE_THRESHOLD,
+                    hybrid_search=rag_app.state.config.ENABLE_RAG_HYBRID_SEARCH,
                 )
                 del data["docs"]
 
@@ -211,15 +238,15 @@ async def check_url(request: Request, call_next):
     return response
 
 
-@app.on_event("startup")
-async def on_startup():
-    if ENABLE_LITELLM:
-        asyncio.create_task(start_litellm_background())
+@app.middleware("http")
+async def update_embedding_function(request: Request, call_next):
+    response = await call_next(request)
+    if "/embedding/update" in request.url.path:
+        webui_app.state.EMBEDDING_FUNCTION = rag_app.state.EMBEDDING_FUNCTION
+    return response
 
 
-app.mount("/api/v1", webui_app)
 app.mount("/litellm/api", litellm_app)
-
 app.mount("/ollama", ollama_app)
 app.mount("/openai/api", openai_app)
 
@@ -227,6 +254,10 @@ app.mount("/images/api/v1", images_app)
 app.mount("/audio/api/v1", audio_app)
 app.mount("/rag/api/v1", rag_app)
 
+app.mount("/api/v1", webui_app)
+
+webui_app.state.EMBEDDING_FUNCTION = rag_app.state.EMBEDDING_FUNCTION
+
 
 @app.get("/api/config")
 async def get_app_config():
@@ -243,9 +274,9 @@ async def get_app_config():
         "version": VERSION,
         "auth": WEBUI_AUTH,
         "default_locale": default_locale,
-        "images": images_app.state.ENABLED,
-        "default_models": webui_app.state.DEFAULT_MODELS,
-        "default_prompt_suggestions": webui_app.state.DEFAULT_PROMPT_SUGGESTIONS,
+        "images": images_app.state.config.ENABLED,
+        "default_models": webui_app.state.config.DEFAULT_MODELS,
+        "default_prompt_suggestions": webui_app.state.config.DEFAULT_PROMPT_SUGGESTIONS,
         "trusted_header_auth": bool(webui_app.state.AUTH_TRUSTED_EMAIL_HEADER),
         "admin_export_enabled": ENABLE_ADMIN_EXPORT,
     }
@@ -254,8 +285,8 @@ async def get_app_config():
 @app.get("/api/config/model/filter")
 async def get_model_filter_config(user=Depends(get_admin_user)):
     return {
-        "enabled": app.state.ENABLE_MODEL_FILTER,
-        "models": app.state.MODEL_FILTER_LIST,
+        "enabled": app.state.config.ENABLE_MODEL_FILTER,
+        "models": app.state.config.MODEL_FILTER_LIST,
     }
 
 
@@ -268,28 +299,28 @@ class ModelFilterConfigForm(BaseModel):
 async def update_model_filter_config(
     form_data: ModelFilterConfigForm, user=Depends(get_admin_user)
 ):
-    app.state.ENABLE_MODEL_FILTER = form_data.enabled
-    app.state.MODEL_FILTER_LIST = form_data.models
+    app.state.config.ENABLE_MODEL_FILTER = form_data.enabled
+    app.state.config.MODEL_FILTER_LIST = form_data.models
 
-    ollama_app.state.ENABLE_MODEL_FILTER = app.state.ENABLE_MODEL_FILTER
-    ollama_app.state.MODEL_FILTER_LIST = app.state.MODEL_FILTER_LIST
+    ollama_app.state.config.ENABLE_MODEL_FILTER = app.state.config.ENABLE_MODEL_FILTER
+    ollama_app.state.config.MODEL_FILTER_LIST = app.state.config.MODEL_FILTER_LIST
 
-    openai_app.state.ENABLE_MODEL_FILTER = app.state.ENABLE_MODEL_FILTER
-    openai_app.state.MODEL_FILTER_LIST = app.state.MODEL_FILTER_LIST
+    openai_app.state.config.ENABLE_MODEL_FILTER = app.state.config.ENABLE_MODEL_FILTER
+    openai_app.state.config.MODEL_FILTER_LIST = app.state.config.MODEL_FILTER_LIST
 
-    litellm_app.state.ENABLE_MODEL_FILTER = app.state.ENABLE_MODEL_FILTER
-    litellm_app.state.MODEL_FILTER_LIST = app.state.MODEL_FILTER_LIST
+    litellm_app.state.ENABLE_MODEL_FILTER = app.state.config.ENABLE_MODEL_FILTER
+    litellm_app.state.MODEL_FILTER_LIST = app.state.config.MODEL_FILTER_LIST
 
     return {
-        "enabled": app.state.ENABLE_MODEL_FILTER,
-        "models": app.state.MODEL_FILTER_LIST,
+        "enabled": app.state.config.ENABLE_MODEL_FILTER,
+        "models": app.state.config.MODEL_FILTER_LIST,
     }
 
 
 @app.get("/api/webhook")
 async def get_webhook_url(user=Depends(get_admin_user)):
     return {
-        "url": app.state.WEBHOOK_URL,
+        "url": app.state.config.WEBHOOK_URL,
     }
 
 
@@ -299,12 +330,12 @@ class UrlForm(BaseModel):
 
 @app.post("/api/webhook")
 async def update_webhook_url(form_data: UrlForm, user=Depends(get_admin_user)):
-    app.state.WEBHOOK_URL = form_data.url
+    app.state.config.WEBHOOK_URL = form_data.url
 
-    webui_app.state.WEBHOOK_URL = app.state.WEBHOOK_URL
+    webui_app.state.WEBHOOK_URL = app.state.config.WEBHOOK_URL
 
     return {
-        "url": app.state.WEBHOOK_URL,
+        "url": app.state.config.WEBHOOK_URL,
     }
 
 
@@ -368,6 +399,11 @@ async def get_opensearch_xml():
     return Response(content=xml_content, media_type="application/xml")
 
 
+@app.get("/health")
+async def healthcheck():
+    return {"status": True}
+
+
 app.mount("/static", StaticFiles(directory=STATIC_DIR), name="static")
 app.mount("/cache", StaticFiles(directory=CACHE_DIR), name="cache")
 
@@ -381,9 +417,3 @@ else:
     log.warning(
         f"Frontend build directory not found at '{FRONTEND_BUILD_DIR}'. Serving API only."
     )
-
-
-@app.on_event("shutdown")
-async def shutdown_event():
-    if ENABLE_LITELLM:
-        await shutdown_litellm_background()

cypress/e2e/chat.cy.ts

@@ -42,5 +42,37 @@ describe('Settings', () => {
 				.find('div[aria-label="Generation Info"]', { timeout: 120_000 }) // Generation Info is created after the stop token is received
 				.should('exist');
 		});
+
+		it('user can share chat', () => {
+			// Click on the model selector
+			cy.get('button[aria-label="Select a model"]').click();
+			// Select the first model
+			cy.get('button[aria-label="model-item"]').first().click();
+			// Type a message
+			cy.get('#chat-textarea').type('Hi, what can you do? A single sentence only please.', {
+				force: true
+			});
+			// Send the message
+			cy.get('button[type="submit"]').click();
+			// User's message should be visible
+			cy.get('.chat-user').should('exist');
+			// Wait for the response
+			cy.get('.chat-assistant', { timeout: 120_000 }) // .chat-assistant is created after the first token is received
+				.find('div[aria-label="Generation Info"]', { timeout: 120_000 }) // Generation Info is created after the stop token is received
+				.should('exist');
+			// spy on requests
+			const spy = cy.spy();
+			cy.intercept('GET', '/api/v1/chats/*', spy);
+			// Open context menu
+			cy.get('#chat-context-menu-button').click();
+			// Click share button
+			cy.get('#chat-share-button').click();
+			// Check if the share dialog is visible
+			cy.get('#copy-and-share-chat-button').should('exist');
+			cy.wrap({}, { timeout: 5000 }).should(() => {
+				// Check if the request was made twice (once for to replace chat object and once more due to change event)
+				expect(spy).to.be.callCount(2);
+			});
+		});
 	});
 });

cypress/e2e/settings.cy.ts

@@ -15,12 +15,8 @@ describe('Settings', () => {
 		cy.loginAdmin();
 		// Visit the home page
 		cy.visit('/');
-		// Open the sidebar if it is not already open
-		cy.get('[aria-label="Open sidebar"]').then(() => {
-			cy.get('button[id="sidebar-toggle-button"]').click();
-		});
-		// Click on the profile link
-		cy.get('button').contains(adminUser.name).click();
+		// Click on the user menu
+		cy.get('button[aria-label="User Menu"]').click();
 		// Click on the settings link
 		cy.get('button').contains('Settings').click();
 	});

docker-compose.api.yaml

-version: '3.8'
-
 services:
   ollama:
     # Expose Ollama API outside the container stack

docker-compose.data.yaml

-version: '3.8'
-
 services:
   ollama:
     volumes:

docker-compose.gpu.yaml

-version: '3.8'
-
 services:
   ollama:
     # GPU support

docker-compose.yaml

-version: '3.8'
-
 services:
   ollama:
     volumes:

docs/CONTRIBUTING.md

@@ -17,7 +17,7 @@ If your issue or contribution pertains directly to the core Ollama technology, p
 
 ### 🚨 Reporting Issues
 
-Noticed something off? Have an idea? Check our [Issues tab](https://github.com/open-webui/oopen-webui/issues) to see if it's already been reported or suggested. If not, feel free to open a new issue. When reporting an issue, please follow our issue templates. These templates are designed to ensure that all necessary details are provided from the start, enabling us to address your concerns more efficiently.
+Noticed something off? Have an idea? Check our [Issues tab](https://github.com/open-webui/open-webui/issues) to see if it's already been reported or suggested. If not, feel free to open a new issue. When reporting an issue, please follow our issue templates. These templates are designed to ensure that all necessary details are provided from the start, enabling us to address your concerns more efficiently.
 
 > [!IMPORTANT]
 >
@@ -54,7 +54,7 @@ Help us make Open WebUI more accessible by improving documentation, writing tuto
 
 Help us make Open WebUI available to a wider audience. In this section, we'll guide you through the process of adding new translations to the project.
 
-We use JSON files to store translations. You can find the existing translation files in the `src/lib/i18n/locales` directory. Each directory corresponds to a specific language, for example, `en-US` for English (US), `fr-FR` for French (France) and so on. You can refer to [ISO 639 Language Codes][http://www.lingoes.net/en/translator/langcode.htm] to find the appropriate code for a specific language.
+We use JSON files to store translations. You can find the existing translation files in the `src/lib/i18n/locales` directory. Each directory corresponds to a specific language, for example, `en-US` for English (US), `fr-FR` for French (France) and so on. You can refer to [ISO 639 Language Codes](http://www.lingoes.net/en/translator/langcode.htm) to find the appropriate code for a specific language.
 
 To add a new language:
 

package.json

 {
 	"name": "open-webui",
-	"version": "0.1.124",
+	"version": "0.1.125",
 	"private": true,
 	"scripts": {
-		"dev": "vite dev --host",
-		"build": "vite build",
+		"dev": "npm run pyodide:fetch && vite dev --host",
+		"build": "npm run pyodide:fetch && vite build",
 		"preview": "vite preview",
 		"check": "svelte-kit sync && svelte-check --tsconfig ./tsconfig.json",
 		"check:watch": "svelte-kit sync && svelte-check --tsconfig ./tsconfig.json --watch",
@@ -12,10 +12,12 @@
 		"lint:frontend": "eslint . --fix",
 		"lint:types": "npm run check",
 		"lint:backend": "pylint backend/",
-		"format": "prettier --plugin-search-dir --write '**/*.{js,ts,svelte,css,md,html,json}'",
+		"format": "prettier --plugin-search-dir --write \"**/*.{js,ts,svelte,css,md,html,json}\"",
 		"format:backend": "black . --exclude \"/venv/\"",
-		"i18n:parse": "i18next --config i18next-parser.config.ts && prettier --write 'src/lib/i18n/**/*.{js,json}'",
-		"cy:open": "cypress open"
+		"i18n:parse": "i18next --config i18next-parser.config.ts && prettier --write \"src/lib/i18n/**/*.{js,json}\"",
+		"cy:open": "cypress open",
+		"test:frontend": "vitest",
+		"pyodide:fetch": "node scripts/prepare-pyodide.js"
 	},
 	"devDependencies": {
 		"@sveltejs/adapter-auto": "^2.0.0",
@@ -41,10 +43,12 @@
 		"tailwindcss": "^3.3.3",
 		"tslib": "^2.4.1",
 		"typescript": "^5.0.0",
-		"vite": "^4.4.2"
+		"vite": "^4.4.2",
+		"vitest": "^1.6.0"
 	},
 	"type": "module",
 	"dependencies": {
+		"@pyscript/core": "^0.4.32",
 		"@sveltejs/adapter-node": "^1.3.1",
 		"async": "^3.2.5",
 		"bits-ui": "^0.19.7",
@@ -59,6 +63,7 @@
 		"js-sha256": "^0.10.1",
 		"katex": "^0.16.9",
 		"marked": "^9.1.0",
+		"pyodide": "^0.26.0-alpha.4",
 		"svelte-sonner": "^0.3.19",
 		"tippy.js": "^6.3.7",
 		"uuid": "^9.0.1"

scripts/prepare-pyodide.js

+const packages = [
+	'requests',
+	'beautifulsoup4',
+	'numpy',
+	'pandas',
+	'matplotlib',
+	'scikit-learn',
+	'scipy',
+	'regex',
+	'seaborn'
+];
+
+import { loadPyodide } from 'pyodide';
+import { writeFile, copyFile, readdir } from 'fs/promises';
+
+async function downloadPackages() {
+	console.log('Setting up pyodide + micropip');
+	const pyodide = await loadPyodide({
+		packageCacheDir: 'static/pyodide'
+	});
+	await pyodide.loadPackage('micropip');
+	const micropip = pyodide.pyimport('micropip');
+	console.log('Downloading Pyodide packages:', packages);
+	await micropip.install(packages);
+	console.log('Pyodide packages downloaded, freezing into lock file');
+	const lockFile = await micropip.freeze();
+	await writeFile('static/pyodide/pyodide-lock.json', lockFile);
+}
+
+async function copyPyodide() {
+	console.log('Copying Pyodide files into static directory');
+	// Copy all files from node_modules/pyodide to static/pyodide
+	for await (const entry of await readdir('node_modules/pyodide')) {
+		await copyFile(`node_modules/pyodide/${entry}`, `static/pyodide/${entry}`);
+	}
+}
+
+await downloadPackages();
+await copyPyodide();

src/app.css

@@ -83,11 +83,31 @@ select {
 	display: none;
 }
 
-.scrollbar-none:active::-webkit-scrollbar-thumb,
-.scrollbar-none:focus::-webkit-scrollbar-thumb,
-.scrollbar-none:hover::-webkit-scrollbar-thumb {
+.scrollbar-hidden:active::-webkit-scrollbar-thumb,
+.scrollbar-hidden:focus::-webkit-scrollbar-thumb,
+.scrollbar-hidden:hover::-webkit-scrollbar-thumb {
 	visibility: visible;
 }
-.scrollbar-none::-webkit-scrollbar-thumb {
+.scrollbar-hidden::-webkit-scrollbar-thumb {
 	visibility: hidden;
 }
+
+.scrollbar-none::-webkit-scrollbar {
+	display: none; /* for Chrome, Safari and Opera */
+}
+
+.scrollbar-none {
+	-ms-overflow-style: none; /* IE and Edge */
+	scrollbar-width: none; /* Firefox */
+}
+
+input::-webkit-outer-spin-button,
+input::-webkit-inner-spin-button {
+	/* display: none; <- Crashes Chrome on hover */
+	-webkit-appearance: none;
+	margin: 0; /* <-- Apparently some margin are still there even though it's hidden */
+}
+
+input[type='number'] {
+	-moz-appearance: textfield; /* Firefox */
+}

src/app.html

@@ -12,6 +12,7 @@
 			title="Open WebUI"
 			href="/opensearch.xml"
 		/>
+
 		<script>
 			// On page load or when changing themes, best to add inline in `head` to avoid FOUC
 			(() => {

src/lib/apis/memories/index.ts

+import { WEBUI_API_BASE_URL } from '$lib/constants';
+
+export const getMemories = async (token: string) => {
+	let error = null;
+
+	const res = await fetch(`${WEBUI_API_BASE_URL}/memories`, {
+		method: 'GET',
+		headers: {
+			Accept: 'application/json',
+			'Content-Type': 'application/json',
+			authorization: `Bearer ${token}`
+		}
+	})
+		.then(async (res) => {
+			if (!res.ok) throw await res.json();
+			return res.json();
+		})
+		.catch((err) => {
+			error = err.detail;
+			console.log(err);
+			return null;
+		});
+
+	if (error) {
+		throw error;
+	}
+
+	return res;
+};
+
+export const addNewMemory = async (token: string, content: string) => {
+	let error = null;
+
+	const res = await fetch(`${WEBUI_API_BASE_URL}/memories/add`, {
+		method: 'POST',
+		headers: {
+			Accept: 'application/json',
+			'Content-Type': 'application/json',
+			authorization: `Bearer ${token}`
+		},
+		body: JSON.stringify({
+			content: content
+		})
+	})
+		.then(async (res) => {
+			if (!res.ok) throw await res.json();
+			return res.json();
+		})
+		.catch((err) => {
+			error = err.detail;
+			console.log(err);
+			return null;
+		});
+
+	if (error) {
+		throw error;
+	}
+
+	return res;
+};
+
+export const queryMemory = async (token: string, content: string) => {
+	let error = null;
+
+	const res = await fetch(`${WEBUI_API_BASE_URL}/memories/query`, {
+		method: 'POST',
+		headers: {
+			Accept: 'application/json',
+			'Content-Type': 'application/json',
+			authorization: `Bearer ${token}`
+		},
+		body: JSON.stringify({
+			content: content
+		})
+	})
+		.then(async (res) => {
+			if (!res.ok) throw await res.json();
+			return res.json();
+		})
+		.catch((err) => {
+			error = err.detail;
+			console.log(err);
+			return null;
+		});
+
+	if (error) {
+		throw error;
+	}
+
+	return res;
+};
+
+export const deleteMemoryById = async (token: string, id: string) => {
+	let error = null;
+
+	const res = await fetch(`${WEBUI_API_BASE_URL}/memories/${id}`, {
+		method: 'DELETE',
+		headers: {
+			Accept: 'application/json',
+			'Content-Type': 'application/json',
+			authorization: `Bearer ${token}`
+		}
+	})
+		.then(async (res) => {
+			if (!res.ok) throw await res.json();
+			return res.json();
+		})
+		.then((json) => {
+			return json;
+		})
+		.catch((err) => {
+			error = err.detail;
+
+			console.log(err);
+			return null;
+		});
+
+	if (error) {
+		throw error;
+	}
+
+	return res;
+};
+
+export const deleteMemoriesByUserId = async (token: string) => {
+	let error = null;
+
+	const res = await fetch(`${WEBUI_API_BASE_URL}/memories/user`, {
+		method: 'DELETE',
+		headers: {
+			Accept: 'application/json',
+			'Content-Type': 'application/json',
+			authorization: `Bearer ${token}`
+		}
+	})
+		.then(async (res) => {
+			if (!res.ok) throw await res.json();
+			return res.json();
+		})
+		.then((json) => {
+			return json;
+		})
+		.catch((err) => {
+			error = err.detail;
+
+			console.log(err);
+			return null;
+		});
+
+	if (error) {
+		throw error;
+	}
+
+	return res;
+};