Merge pull request #790 from thepetk/ft/add_bearer_auth_support

Add Bearer (token) authentication support to recipes

recipes/audio/audio_to_text/app/whisper_client.py

@@ -8,13 +8,17 @@
 st.markdown("Upload an audio file you wish to have translated")
 endpoint = os.getenv("MODEL_ENDPOINT", default="http://0.0.0.0:8001")
 endpoint = f"{endpoint}/inference"
+endpoint_bearer = os.getenv("MODEL_ENDPOINT_BEARER")
+request_kwargs = {}
+if endpoint_bearer is not None:
+    request_kwargs["headers"] = {"Authorization": f"Bearer {endpoint_bearer}"}
 audio = st.file_uploader("", type=["wav","mp3","mp4","flac"], accept_multiple_files=False)
 # read audio file
 if audio:
     audio_bytes = audio.read()
     st.audio(audio_bytes, format='audio/wav', start_time=0)
-    files = {'file': audio_bytes}
-    response = requests.post(endpoint, files=files)
+    request_kwargs["files"] = {'file': audio_bytes}
+    response = requests.post(endpoint, **request_kwargs)
     response_json = response.json()
     st.subheader(f"Translated Text")
     st.text_area(label="", value=response_json['text'], height=300)

recipes/computer_vision/object_detection/app/object_detection_client.py

@@ -7,8 +7,11 @@
 
 st.title("🕵️‍♀️ Object Detection")
 endpoint =os.getenv("MODEL_ENDPOINT", default = "http://0.0.0.0:8000")
+endpoint_bearer = os.getenv("MODEL_ENDPOINT_BEARER")
 headers = {"accept": "application/json",
            "Content-Type": "application/json"}
+if endpoint_bearer:
+    headers["Authorization"] = f"Bearer {endpoint_bearer}"
 image = st.file_uploader("Upload Image")
 window = st.empty()
 

recipes/natural_language_processing/chatbot/app/chatbot_ui.py

@@ -12,6 +12,10 @@
 model_service = os.getenv("MODEL_ENDPOINT",
                           "http://localhost:8001")
 model_service = f"{model_service}/v1"
+model_service_bearer = os.getenv("MODEL_ENDPOINT_BEARER")
+request_kwargs = {}
+if model_service_bearer is not None:
+    request_kwargs = {"headers": {"Authorization": f"Bearer {model_service_bearer}"}}
 
 @st.cache_resource(show_spinner=False)
 def checking_model_service():
@@ -20,8 +24,8 @@ def checking_model_service():
     ready = False
     while not ready:
         try:
-            request_cpp = requests.get(f'{model_service}/models')
-            request_ollama = requests.get(f'{model_service[:-2]}api/tags')
+            request_cpp = requests.get(f'{model_service}/models', **request_kwargs)
+            request_ollama = requests.get(f'{model_service[:-2]}api/tags', **request_kwargs)
             if request_cpp.status_code == 200:
                 server = "Llamacpp_Python"
                 ready = True
@@ -37,7 +41,7 @@ def checking_model_service():
 
 def get_models():
     try:
-        response = requests.get(f"{model_service[:-2]}api/tags")
+        response = requests.get(f"{model_service[:-2]}api/tags", **request_kwargs)
         return [i["name"].split(":")[0] for i in  
             json.loads(response.content)["models"]]
     except:

recipes/natural_language_processing/codegen/app/codegen-app.py

@@ -10,6 +10,10 @@
 
 model_service = os.getenv("MODEL_ENDPOINT", "http://localhost:8001")
 model_service = f"{model_service}/v1"
+model_service_bearer = os.getenv("MODEL_ENDPOINT_BEARER")
+request_kwargs = {}
+if model_service_bearer is not None:
+    request_kwargs = {"headers": {"Authorization": f"Bearer {model_service_bearer}"}}
 
 @st.cache_resource(show_spinner=False)
 def checking_model_service():
@@ -18,7 +22,7 @@ def checking_model_service():
     ready = False
     while not ready:
         try:
-            request = requests.get(f'{model_service}/models')
+            request = requests.get(f'{model_service}/models', **request_kwargs)
             if request.status_code == 200:
                 ready = True
         except:

recipes/natural_language_processing/rag/app/rag_app.py

@@ -12,6 +12,7 @@
 
 model_service = os.getenv("MODEL_ENDPOINT","http://0.0.0.0:8001")
 model_service = f"{model_service}/v1"
+model_service_bearer = os.getenv("MODEL_ENDPOINT_BEARER")
 model_name = os.getenv("MODEL_NAME", "") 
 chunk_size = os.getenv("CHUNK_SIZE", 150)
 embedding_model = os.getenv("EMBEDDING_MODEL","BAAI/bge-base-en-v1.5")
@@ -75,7 +76,7 @@ def read_file(file):
 
 
 llm = ChatOpenAI(base_url=model_service, 
-                 api_key="EMPTY",
+                 api_key="EMPTY" if model_service_bearer is None else model_service_bearer,
                  model=model_name,
                  streaming=True,
                  callbacks=[StreamlitCallbackHandler(st.container(),

recipes/natural_language_processing/summarizer/app/summarizer.py

@@ -15,6 +15,10 @@
 model_service = os.getenv("MODEL_ENDPOINT",
                           "http://localhost:8001")
 model_service = f"{model_service}/v1"
+model_service_bearer = os.getenv("MODEL_ENDPOINT_BEARER")
+request_kwargs = {}
+if model_service_bearer is not None:
+    request_kwargs["headers"] = {"Authorization": f"Bearer {model_service_bearer}"}
 
 @st.cache_resource(show_spinner=False)
 def checking_model_service():
@@ -23,7 +27,7 @@ def checking_model_service():
     ready = False
     while not ready:
         try:
-            request = requests.get(f'{model_service}/models')
+            request = requests.get(f'{model_service}/models', **request_kwargs)
             if request.status_code == 200:
                 ready = True
         except:
@@ -53,8 +57,8 @@ def chunk_text(text):
     text_chunks = text_splitter.create_documents([text])
     for chunk in text_chunks:
         chunk = chunk.page_content
-        count = requests.post(f"{model_service[:-2]}extras/tokenize/count",
-                  json={"input":chunk}).content
+        chunk_kwargs = request_kwargs | {"json": {"input": chunk}}
+        count = requests.post(f"{model_service[:-2]}/v1/extras/tokenize/count", **chunk_kwargs).content
         count = json.loads(count)["count"]
         if count >= 2048:
             split_append_chunk(chunk, chunks)