Add Retry Logic to http requests

.gitignore

@@ -12,6 +12,7 @@ __pycache__/
 *.pyc
 *.pyo
 .python-version
+api/api.egg-info/*
 
 # Visual Studio Code
 .vscode/

agent/agent/chain.py

@@ -40,7 +40,7 @@ def template(self) -> dict[str, str]:
         system = (
             {
                 "role": "system",
-                "content": f"""You are Bloom, a subversive-minded learning companion. Your job is to employ your theory of mind skills to predict the user’s mental state.
+                "content": f"""You are Bloom, a subversive-minded learning companion. Your job is to employ your theory of mind skills to predict the user's mental state.
     Generate a thought that makes a prediction about the user's needs given current dialogue and also lists other pieces of data that would help improve your prediction
     previous commentary: {self.history}""",
             },

api/pyproject.toml

@@ -2,9 +2,7 @@
 name = "api"
 version = "0.6.0"
 description = "The REST API Implementation of Tutor-GPT"
-authors = [
-    {name = "Plastic Labs", email = "[email protected]"},
-]
+authors = [{ name = "Plastic Labs", email = "[email protected]" }]
 requires-python = ">=3.11"
 dependencies = [
     "fastapi[standard]>=0.112.2",
@@ -16,4 +14,4 @@ dependencies = [
 
 [tool.uv.sources]
 # agent = { path = "../agent", editable = true }
-agent = {workspace=true}
+agent = { workspace = true }

api/routers/chat.py

@@ -1,128 +1,130 @@
-from typing import Optional
-from fastapi import APIRouter, HTTPException, Body
-from fastapi.responses import StreamingResponse
-from pydantic import BaseModel
+from fastapi import APIRouter, HTTPException
+from fastapi.responses import StreamingResponse, JSONResponse
 
 from api import schemas
 from api.dependencies import app, honcho
 
 from agent.chain import ThinkCall, RespondCall
 
+import logging
+
 router = APIRouter(prefix="/api", tags=["chat"])
 
 
 @router.post("/stream")
-async def stream(
-    inp: schemas.ConversationInput,
+async def stream(inp: schemas.ConversationInput):
+    try:
+        user = honcho.apps.users.get_or_create(app_id=app.id, name=inp.user_id)
+
+        async def convo_turn():
+            thought = ""
+            response = ""
+            try:
+                thought_stream = ThinkCall(
+                    user_input=inp.message,
+                    app_id=app.id,
+                    user_id=user.id,
+                    session_id=str(inp.conversation_id),
+                    honcho=honcho,
+                ).stream()
+                for chunk in thought_stream:
+                    thought += chunk
+                    yield chunk
+
+                yield "❀"
+                response_stream = RespondCall(
+                    user_input=inp.message,
+                    thought=thought,
+                    app_id=app.id,
+                    user_id=user.id,
+                    session_id=str(inp.conversation_id),
+                    honcho=honcho,
+                ).stream()
+                for chunk in response_stream:
+                    response += chunk
+                    yield chunk
+                yield "❀"
+            except Exception as e:
+                logging.error(f"Error during streaming: {str(e)}")
+                yield f"Error: {str(e)}"
+                return
+
+            await create_messages_and_metamessages(
+                app.id, user.id, inp.conversation_id, inp.message, thought, response
+            )
+
+        return StreamingResponse(convo_turn())
+    except Exception as e:
+        logging.error(f"An error occurred: {str(e)}")
+        if "rate limit" in str(e).lower():
+            return JSONResponse(
+                status_code=429,
+                content={
+                    "error": "rate_limit_exceeded",
+                    "message": "Rate limit exceeded. Please try again later.",
+                },
+            )
+        else:
+            return JSONResponse(
+                status_code=500,
+                content={
+                    "error": "internal_server_error",
+                    "message": "An internal server error has occurred.",
+                },
+            )
+
+
+async def create_messages_and_metamessages(
+    app_id, user_id, conversation_id, user_message, thought, ai_response
 ):
-    """Stream the response too the user, currently only used by the Web UI and has integration to be able to use Honcho is not anonymous"""
-    user = honcho.apps.users.get_or_create(app_id=app.id, name=inp.user_id)
-
-    def convo_turn():
-        thought_stream = ThinkCall(
-            user_input=inp.message,
-            app_id=app.id,
-            user_id=user.id,
-            session_id=str(inp.conversation_id),
-            honcho=honcho,
-        ).stream()
-        thought = ""
-        for chunk in thought_stream:
-            thought += chunk
-            yield chunk
-
-        yield "❀"
-        response_stream = RespondCall(
-            user_input=inp.message,
-            thought=thought,
-            app_id=app.id,
-            user_id=user.id,
-            session_id=str(inp.conversation_id),
-            honcho=honcho,
-        ).stream()
-        response = ""
-        for chunk in response_stream:
-            response += chunk
-            yield chunk
-        yield "❀"
-
-        honcho.apps.users.sessions.messages.create(
+    try:
+        # These operations will use the DB layer's built-in retry logic
+        await honcho.apps.users.sessions.messages.create(
             is_user=True,
-            session_id=str(inp.conversation_id),
-            app_id=app.id,
-            user_id=user.id,
-            content=inp.message,
+            session_id=str(conversation_id),
+            app_id=app_id,
+            user_id=user_id,
+            content=user_message,
         )
-        new_ai_message = honcho.apps.users.sessions.messages.create(
+        new_ai_message = await honcho.apps.users.sessions.messages.create(
             is_user=False,
-            session_id=str(inp.conversation_id),
-            app_id=app.id,
-            user_id=user.id,
-            content=response,
+            session_id=str(conversation_id),
+            app_id=app_id,
+            user_id=user_id,
+            content=ai_response,
         )
-        honcho.apps.users.sessions.metamessages.create(
-            app_id=app.id,
-            session_id=str(inp.conversation_id),
-            user_id=user.id,
+        await honcho.apps.users.sessions.metamessages.create(
+            app_id=app_id,
+            session_id=str(conversation_id),
+            user_id=user_id,
             message_id=new_ai_message.id,
             metamessage_type="thought",
             content=thought,
         )
-    return StreamingResponse(convo_turn())
+    except Exception as e:
+        logging.error(f"Error in create_messages_and_metamessages: {str(e)}")
+        raise  # Re-raise the exception to be handled by the caller
+
 
 @router.get("/thought/{message_id}")
 async def get_thought(conversation_id: str, message_id: str, user_id: str):
-    user = honcho.apps.users.get_or_create(app_id=app.id, name=user_id)
-    thought = honcho.apps.users.sessions.metamessages.list(
-        session_id=conversation_id,
-        app_id=app.id,
-        user_id=user.id,
-        message_id=message_id,
-        metamessage_type="thought"
-    )
-    # In practice, there should only be one thought per message
-    return {"thought": thought.items[0].content if thought.items else None}
-
-class ReactionBody(BaseModel):
-    reaction: Optional[str] = None
-
-@router.post("/reaction/{message_id}")
-async def add_or_remove_reaction(
-    conversation_id: str,
-    message_id: str,
-    user_id: str,
-    body: ReactionBody
-):
-    reaction = body.reaction
-
-    if reaction is not None and reaction not in ["thumbs_up", "thumbs_down"]:
-        raise HTTPException(status_code=400, detail="Invalid reaction type")
-
-    user = honcho.apps.users.get_or_create(app_id=app.id, name=user_id)
-
-    message = honcho.apps.users.sessions.messages.get(
-        app_id=app.id,
-        session_id=conversation_id,
-        user_id=user.id,
-        message_id=message_id
-    )
-
-    if not message:
-        raise HTTPException(status_code=404, detail="Message not found")
-
-    metadata = message.metadata or {}
-
-    if reaction is None:
-        metadata.pop('reaction', None)
-    else:
-        metadata['reaction'] = reaction
-
-    honcho.apps.users.sessions.messages.update(
-        app_id=app.id,
-        session_id=conversation_id,
-        user_id=user.id,
-        message_id=message_id,
-        metadata=metadata
-    )
-
-    return {"status": "Reaction updated successfully"}
+    try:
+        user = honcho.apps.users.get_or_create(app_id=app.id, name=user_id)
+        thought = honcho.apps.users.sessions.metamessages.list(
+            session_id=conversation_id,
+            app_id=app.id,
+            user_id=user.id,
+            message_id=message_id,
+            metamessage_type="thought",
+        )
+        # In practice, there should only be one thought per message
+        return {"thought": thought.items[0].content if thought.items else None}
+    except Exception as e:
+        logging.error(f"An error occurred: {str(e)}")
+        return JSONResponse(
+            status_code=500,
+            content={
+                "error": "internal_server_error",
+                "message": "An internal server error has occurred.",
+            },
+        )

www/components/messagebox.tsx

@@ -22,6 +22,8 @@ interface MessageBoxProps {
   onReactionAdded: (messageId: string, reaction: Reaction) => Promise<void>;
 }
 
+export type Reaction = "thumbs_up" | "thumbs_down" | null;
+
 export default function MessageBox({
   isUser,
   userId,

www/package.json

@@ -26,6 +26,7 @@
     "react-toggle-dark-mode": "^1.1.1",
     "rehype-katex": "^7.0.1",
     "remark-math": "^6.0.0",
+    "retry": "^0.13.1",
     "sharp": "^0.32.6",
     "stripe": "^16.12.0",
     "sweetalert2": "^11.14.2",
@@ -38,6 +39,7 @@
     "@types/react": "18.2.21",
     "@types/react-dom": "18.2.7",
     "@types/react-syntax-highlighter": "^15.5.13",
+    "@types/retry": "^0.12.5",
     "@types/uuid": "^9.0.8",
     "autoprefixer": "10.4.15",
     "encoding": "^0.1.13",