app.py

import os
from fastapi import FastAPI, UploadFile
from doctr.models import ocr_predictor
from doctr.io import DocumentFile

# Set DocTR cache directory before importing models
os.environ["DOCTR_CACHE_DIR"] = "/app/.cache"
os.makedirs("/app/.cache", exist_ok=True)

app = FastAPI()

# Initialize DocTR OCR model once at startup
model = ocr_predictor(pretrained=True)

@app.post("/ocr")
async def extract_text(file: UploadFile):
    # Save uploaded file temporarily
    image_path = "temp.png"
    with open(image_path, "wb") as f:
        f.write(await file.read())

    # Read document and run OCR
    doc = DocumentFile.from_images(image_path)
    result = model(doc)

    # Extract all words as a single string
    text = " ".join([
        word.value
        for page in result.pages
        for block in page.blocks
        for line in block.lines
        for word in line.words
    ])

    return {"text": text}