Better envs

.backend_env.example

@@ -1,4 +1,4 @@
 SUPABASE_URL="XXXXX"
 SUPABASE_SERVICE_KEY="eyXXXXX"
 OPENAI_API_KEY="sk-XXXXXX"
-anthropic_api_key="XXXXXX"
+ANTHROPIC_API_KEY="XXXXXX"

.frontend_env.example

@@ -1 +1,2 @@
 ENV=local
+NEXT_PUBLIC_BACKEND_URL=http://localhost:5000

.gitignore

@@ -3,6 +3,7 @@ secondbrain/
 .streamlit/secrets.toml
 **/*.pyc
 toto.txt
+*.ipynb
 
 
 

README.md

@@ -81,11 +81,11 @@ Additionally, you'll need a [Supabase](https://supabase.com/) account for:
 - **Step 2**: Copy the `.XXXXX_env` files
 
 ```bash
-cp .backend_env.example .backend_env
-cp .frontend_env.example .frontend_env
+cp .backend_env.example backend/.env
+cp .frontend_env.example frontend/.env
 ```
 
-- **Step 3**: Update the `.backend_env` file 
+- **Step 3**: Update the `backend/.env` file 
 
 > _Your `supabase_service_key` can be found in your Supabase dashboard under Project Settings -> API. Use the `anon` `public` key found in the `Project API keys` section._
 
@@ -95,6 +95,8 @@ cp .frontend_env.example .frontend_env
 
 [Migration Script 2](scripts/supabase_usage_table.sql)
 
+[Migration Script 3](scripts/supabase_vector_store_document.sql)
+
 - **Step 5**: Launch the app
 
 ```bash

backend/Dockerfile

@@ -8,4 +8,4 @@ RUN pip install --no-cache-dir --upgrade -r /code/requirements.txt --timeout 100
 
 COPY . /code/
 
-CMD ["uvicorn", "api:app", "--reload", "--host", "0.0.0.0", "--port", "5000"]
+CMD ["uvicorn", "api:app", "--reload", "--host", "0.0.0.0", "--port", "5050"]

backend/api.py

@@ -67,14 +67,12 @@ memory = ConversationBufferMemory(
 class ChatMessage(BaseModel):
     model: str = "gpt-3.5-turbo"
     question: str
-    history: List[Tuple[str, str]]  # A list of tuples where each tuple is (speaker, text)
+    # A list of tuples where each tuple is (speaker, text)
+    history: List[Tuple[str, str]]
     temperature: float = 0.0
     max_tokens: int = 256
 
 
-
-
-
 file_processors = {
     ".txt": process_txt,
     ".csv": process_csv,
@@ -95,6 +93,7 @@ file_processors = {
     ".ipynb": process_ipnyb,
 }
 
+
 async def filter_file(file: UploadFile, supabase, vector_store, stats_db):
     if await file_already_exists(supabase, file):
         return {"message": f"🤔 {file.filename} already exists.", "type": "warning"}
@@ -108,17 +107,19 @@ async def filter_file(file: UploadFile, supabase, vector_store, stats_db):
         else:
             return {"message": f"❌ {file.filename} is not supported.", "type": "error"}
 
+
 @app.post("/upload")
 async def upload_file(file: UploadFile):
     message = await filter_file(file, supabase, vector_store, stats_db=None)
     return message
 
+
 @app.post("/chat/")
 async def chat_endpoint(chat_message: ChatMessage):
     history = chat_message.history
     # Logic from your Streamlit app goes here. For example:
 
-    #this overwrites the built-in prompt of the ConversationalRetrievalChain
+    # this overwrites the built-in prompt of the ConversationalRetrievalChain
     ConversationalRetrievalChain.prompts = LANGUAGE_PROMPT
 
     qa = None
@@ -137,6 +138,7 @@ async def chat_endpoint(chat_message: ChatMessage):
 
     return {"history": history}
 
+
 @app.post("/crawl/")
 async def crawl_endpoint(crawl_website: CrawlWebsite):
 
@@ -152,9 +154,11 @@ async def crawl_endpoint(crawl_website: CrawlWebsite):
     message = await filter_file(file, supabase, vector_store, stats_db=None)
     return message
 
+
 @app.get("/explore")
 async def explore_endpoint():
-    response = supabase.table("documents").select("name:metadata->>file_name, size:metadata->>file_size", count="exact").execute()
+    response = supabase.table("documents").select(
+        "name:metadata->>file_name, size:metadata->>file_size", count="exact").execute()
     documents = response.data  # Access the data from the response
     # Convert each dictionary to a tuple of items, then to a set to remove duplicates, and then back to a dictionary
     unique_data = [dict(t) for t in set(tuple(d.items()) for d in documents)]
@@ -163,22 +167,23 @@ async def explore_endpoint():
 
     return {"documents": unique_data}
 
+
 @app.delete("/explore/{file_name}")
 async def delete_endpoint(file_name: str):
-    response = supabase.table("documents").delete().match({"metadata->>file_name": file_name}).execute()
+    response = supabase.table("documents").delete().match(
+        {"metadata->>file_name": file_name}).execute()
     return {"message": f"{file_name} has been deleted."}
 
+
 @app.get("/explore/{file_name}")
 async def download_endpoint(file_name: str):
-    response = supabase.table("documents").select("metadata->>file_name, metadata->>file_size, metadata->>file_extension, metadata->>file_url").match({"metadata->>file_name": file_name}).execute()
+    response = supabase.table("documents").select(
+        "metadata->>file_name, metadata->>file_size, metadata->>file_extension, metadata->>file_url").match({"metadata->>file_name": file_name}).execute()
     documents = response.data
-    ### Returns all documents with the same file name
+    # Returns all documents with the same file name
     return {"documents": documents}
 
 
-
 @app.get("/")
 async def root():
     return {"message": "Hello World"}
-
-

docker-compose.yml

@@ -3,7 +3,7 @@ version: "3"
 services:
   frontend:
     env_file:
-      - .frontend_env
+      - ./frontend/.env
     build:
       context: frontend
       dockerfile: Dockerfile
@@ -17,9 +17,7 @@ services:
       - 3000:3000
   backend:
     env_file:
-      - .backend_env
-    environment:
-      - supabase_url="totot"
+      - ./backend/.env
     build:
       context: backend
       dockerfile: Dockerfile
@@ -28,4 +26,4 @@ services:
     volumes:
       - ./backend/:/code/
     ports:
-      - 5000:5000
+      - 5050:5050

frontend/.env.example

@@ -1 +1,2 @@
 ENV=local
+BACKEND_URL="http://localhost:5050"

frontend/app/chat/page.tsx

@@ -28,7 +28,7 @@ export default function ChatPage() {
   const askQuestion = async () => {
     setHistory((hist) => [...hist, ["user", question]]);
     setIsPending(true);
-    const response = await axios.post("http://localhost:5000/chat/", {
+    const response = await axios.post(`${process.env.NEXT_PUBLIC_BACKEND_URL}/chat/`, {
       model,
       question,
       history,

frontend/app/explore/page.tsx

@@ -18,7 +18,8 @@ export default function ExplorePage() {
 
     const fetchDocuments = async () => {
         try {
-            const response = await axios.get<{ documents: Document[] }>('http://localhost:5000/explore');
+            console.log(`Fetching documents from ${process.env.NEXT_PUBLIC_BACKEND_URL}/explore`);
+            const response = await axios.get<{ documents: Document[] }>(`${process.env.NEXT_PUBLIC_BACKEND_URL}/explore`);
             setDocuments(response.data.documents);
         } catch (error) {
             console.error('Error fetching documents', error);

frontend/app/upload/page.tsx

@@ -61,7 +61,7 @@ export default function UploadPage() {
     formData.append("file", file);
     try {
       const response = await axios.post(
-        "http://localhost:5000/upload",
+        `${process.env.NEXT_PUBLIC_BACKEND_URL}/upload`,
         formData
       );
 

frontend/package.json

@@ -34,6 +34,7 @@
   },
   "devDependencies": {
     "@tailwindcss/typography": "^0.5.9",
+    "@types/next": "^9.0.0",
     "react-icons": "^4.8.0"
   }
 }

scripts/supabase_vector_store_document.sql

@@ -1,7 +1,7 @@
 create extension vector;
 
 -- Create a table to store your documents
-create table documents (
+create table if not exists documents (
 id bigserial primary key,
 content text, -- corresponds to Document.pageContent
 metadata jsonb, -- corresponds to Document.metadata

scripts/supabase_vector_store_summary.sql

@@ -0,0 +1,38 @@
+-- Create a table to store your summaries
+create table if not exists summaries (
+    id bigserial primary key,
+    document_id bigint references documents(id),
+    content text, -- corresponds to the summarized content
+    metadata jsonb, -- corresponds to Document.metadata
+    embedding vector(1536) -- 1536 works for OpenAI embeddings, change if needed
+);
+
+CREATE OR REPLACE FUNCTION match_summaries(query_embedding vector(1536), match_count int, match_threshold float)
+    RETURNS TABLE(
+        id bigint,
+        document_id bigint,
+        content text,
+        metadata jsonb,
+        -- we return matched vectors to enable maximal marginal relevance searches
+        embedding vector(1536),
+        similarity float)
+    LANGUAGE plpgsql
+    AS $$
+    # variable_conflict use_column
+BEGIN
+    RETURN query
+    SELECT
+        id,
+        document_id,
+        content,
+        metadata,
+        embedding,
+        1 -(summaries.embedding <=> query_embedding) AS similarity
+    FROM
+        summaries
+    WHERE 1 - (summaries.embedding <=> query_embedding) > match_threshold
+    ORDER BY
+        summaries.embedding <=> query_embedding
+    LIMIT match_count;
+END;
+$$;