# pyrefly: ignore [missing-import] from dotenv import load_dotenv load_dotenv() import os # pyrefly: ignore [missing-import] from fastapi import FastAPI, HTTPException # pyrefly: ignore [missing-import] from fastapi.middleware.cors import CORSMiddleware # pyrefly: ignore [missing-import] from pydantic import BaseModel, Field # Import core RAG logic functions import core_logic app = FastAPI( title="AI Browser RAG Backend", description="Backend API for text parsing, embedding, retrieval, LLM querying, and AI actions", version="2.0.0" ) # Allow CORS for all origins since it will be deployed app.add_middleware( CORSMiddleware, allow_origins=["*"], allow_credentials=True, allow_methods=["*"], allow_headers=["*"], ) # --- Pydantic Schemas --- class ProcessHtmlRequest(BaseModel): html: str chunk_size: int = Field(default=50, ge=1) overlap: int = Field(default=10, ge=0) class ProcessHtmlResponse(BaseModel): chunks: list[str] class EmbedRequest(BaseModel): texts: list[str] class EmbedResponse(BaseModel): embeddings: list[list[float]] class RetrieveRequest(BaseModel): query: str chunks: list[str] embeddings: list[list[float]] top_k: int = Field(default=2, ge=1) class RetrievalResultItem(BaseModel): score: float index: int text: str class RetrieveResponse(BaseModel): results: list[RetrievalResultItem] class AnswerRequest(BaseModel): query: str context: str api_key: str | None = None class AnswerResponse(BaseModel): answer: str class UnifiedRagRequest(BaseModel): html: str query: str chunk_size: int = Field(default=50, ge=1) overlap: int = Field(default=10, ge=0) top_k: int = Field(default=2, ge=1) title: str = "" url: str = "" selected_text: str = "" api_key: str | None = None class UnifiedRagResponse(BaseModel): answer: str retrieved_chunks: list[RetrievalResultItem] # --- AI Action Schemas --- class TextRequest(BaseModel): text: str api_key: str | None = None class TranslateRequest(BaseModel): text: str target_language: str api_key: str | None = None class ExtractRequest(BaseModel): text: str data_type: str api_key: str | None = None class GeneratePostRequest(BaseModel): text: str platform: str api_key: str | None = None class RewriteRequest(BaseModel): text: str style: str api_key: str | None = None class ActionResponse(BaseModel): result: str # --- Route Handlers --- @app.get("/") def read_root(): return {"status": "ok", "message": "AI Browser RAG Backend API is running.", "version": "2.0.0"} @app.post("/process-html", response_model=ProcessHtmlResponse) def process_html(payload: ProcessHtmlRequest): try: clean_text = core_logic.extract_clean_prose(payload.html) chunks = core_logic.chunk_text(clean_text, payload.chunk_size, payload.overlap) return ProcessHtmlResponse(chunks=chunks) except Exception as e: raise HTTPException(status_code=500, detail=str(e)) @app.post("/embed", response_model=EmbedResponse) def embed_chunks(payload: EmbedRequest): try: embeddings = core_logic.embed_texts(payload.texts) return EmbedResponse(embeddings=embeddings) except Exception as e: raise HTTPException(status_code=500, detail=str(e)) @app.post("/retrieve", response_model=RetrieveResponse) def retrieve(payload: RetrieveRequest): try: if len(payload.chunks) != len(payload.embeddings): raise HTTPException(status_code=400, detail="The number of chunks must match the number of embeddings.") results = core_logic.retrieve_top_chunks(payload.query, payload.chunks, payload.embeddings, payload.top_k) return RetrieveResponse(results=results) except HTTPException as he: raise he except Exception as e: raise HTTPException(status_code=500, detail=str(e)) @app.post("/answer", response_model=AnswerResponse) def get_answer(payload: AnswerRequest): try: answer = core_logic.query_llm(payload.query, payload.context, payload.api_key) return AnswerResponse(answer=answer) except Exception as e: raise HTTPException(status_code=500, detail=str(e)) @app.post("/rag", response_model=UnifiedRagResponse) def run_unified_rag(payload: UnifiedRagRequest): try: # Step 1: Clean and chunk HTML clean_text = core_logic.extract_clean_prose(payload.html) chunks = core_logic.chunk_text(clean_text, payload.chunk_size, payload.overlap) if not chunks: return UnifiedRagResponse(answer="No text chunks could be extracted from the HTML content.", retrieved_chunks=[]) # Step 2: Embed chunks embeddings = core_logic.embed_texts(chunks) # Step 3: Retrieve top chunks top_chunks = core_logic.retrieve_top_chunks(payload.query, chunks, embeddings, payload.top_k) # Step 4: Format context and query LLM context = "\n\n".join([f"[Source Chunk {item['index'] + 1}]: {item['text']}" for item in top_chunks]) answer = core_logic.query_llm( query=payload.query, context=context, title=payload.title, url=payload.url, selected_text=payload.selected_text, api_key=payload.api_key ) return UnifiedRagResponse(answer=answer, retrieved_chunks=top_chunks) except Exception as e: raise HTTPException(status_code=500, detail=str(e)) # --- AI Action Endpoints --- @app.post("/summarize", response_model=ActionResponse) def summarize(payload: TextRequest): try: result = core_logic.summarize_text(payload.text, payload.api_key) return ActionResponse(result=result) except Exception as e: raise HTTPException(status_code=500, detail=str(e)) @app.post("/explain", response_model=ActionResponse) def explain(payload: TextRequest): try: result = core_logic.explain_text(payload.text, payload.api_key) return ActionResponse(result=result) except Exception as e: raise HTTPException(status_code=500, detail=str(e)) @app.post("/translate", response_model=ActionResponse) def translate(payload: TranslateRequest): try: result = core_logic.translate_text(payload.text, payload.target_language, payload.api_key) return ActionResponse(result=result) except Exception as e: raise HTTPException(status_code=500, detail=str(e)) @app.post("/extract", response_model=ActionResponse) def extract(payload: ExtractRequest): try: result = core_logic.extract_data(payload.text, payload.data_type, payload.api_key) return ActionResponse(result=result) except Exception as e: raise HTTPException(status_code=500, detail=str(e)) @app.post("/generate-post", response_model=ActionResponse) def generate_post(payload: GeneratePostRequest): try: result = core_logic.generate_post(payload.text, payload.platform, payload.api_key) return ActionResponse(result=result) except Exception as e: raise HTTPException(status_code=500, detail=str(e)) @app.post("/rewrite", response_model=ActionResponse) def rewrite(payload: RewriteRequest): try: result = core_logic.rewrite_text(payload.text, payload.style, payload.api_key) return ActionResponse(result=result) except Exception as e: raise HTTPException(status_code=500, detail=str(e)) @app.post("/key-points", response_model=ActionResponse) def key_points(payload: TextRequest): try: result = core_logic.list_key_points(payload.text, payload.api_key) return ActionResponse(result=result) except Exception as e: raise HTTPException(status_code=500, detail=str(e)) @app.post("/faqs", response_model=ActionResponse) def faqs(payload: TextRequest): try: result = core_logic.generate_faqs(payload.text, payload.api_key) return ActionResponse(result=result) except Exception as e: raise HTTPException(status_code=500, detail=str(e)) @app.post("/study-notes", response_model=ActionResponse) def study_notes(payload: TextRequest): try: result = core_logic.create_study_notes(payload.text, payload.api_key) return ActionResponse(result=result) except Exception as e: raise HTTPException(status_code=500, detail=str(e)) if __name__ == "__main__": # pyrefly: if __name__ == "__main__": import uvicorn import os port = int(os.environ.get("PORT", 8000)) uvicorn.run(app, host="0.0.0.0", port=port)