Update app.py

app.py

@@ -1,10 +1,15 @@
+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
+# Initialize DocTR OCR model once at startup
 model = ocr_predictor(pretrained=True)
 
 @app.post("/ocr")
@@ -14,11 +19,11 @@ async def extract_text(file: UploadFile):
     with open(image_path, "wb") as f:
         f.write(await file.read())
 
-    # Read the document
+    # Read document and run OCR
     doc = DocumentFile.from_images(image_path)
     result = model(doc)
 
-    # Extract text from all pages, blocks, lines, words
+    # Extract all words as a single string
     text = " ".join([
         word.value
         for page in result.pages