app.py

from fastapi import FastAPI, File, UploadFile
import pdfplumber
import io

app = FastAPI()

@app.post("/extract")
async def extract(file: UploadFile = File(...)):
    pdf_bytes = await file.read()

    stream = []

    with pdfplumber.open(io.BytesIO(pdf_bytes)) as pdf:
        for page in pdf.pages:

            # Extract text first
            text = page.extract_text()
            if text:
                stream.append({
                    "type": "text",
                    "content": text
                })

            # Extract tables
            tables = page.extract_tables()
            for table in tables:
                stream.append({
                    "type": "table",
                    "content": table
                })

    return {
        "stream": stream
    }

@app.get("/health")
async def health():
    return {
        "status": "ok"
    }