Add files

app.py

@@ -1,3 +1,7 @@
+import warnings
+warnings.filterwarnings("ignore", message=".*pin_memory.*")
+
+
 import os
 import io
 import easyocr
@@ -17,12 +21,12 @@ USER_NET_DIR = os.path.join(MODEL_DIR, "user_network")
 os.makedirs(MODEL_DIR, exist_ok=True)
 os.makedirs(USER_NET_DIR, exist_ok=True)
 
-# ✅ preload reader with cached models
+# ✅ preload reader
 reader = easyocr.Reader(
-    ['en', 'hi'],  # langs (reduce if only English needed)
+    ['en', 'hi'],
     model_storage_directory=MODEL_DIR,
     user_network_directory=USER_NET_DIR,
-    download_enabled=False   # 🚫 block downloads at runtime
+    download_enabled=False
 )
 
 # =========================
@@ -35,22 +39,14 @@ async def root():
     return {"message": "OCR API is running on Hugging Face 🚀"}
 
 
-def run_ocr_on_image(image: Image.Image):
-    """Convert PIL → numpy and run OCR"""
+def run_ocr_on_image(image: Image.Image) -> str:
+    """Convert PIL → numpy, run OCR, return plain text"""
     image_np = np.array(image)
     results = reader.readtext(image_np)
 
-    text_results = []
-    for bbox, text, prob in results:
-        # ✅ convert bbox coords to plain Python floats
-        bbox_py = [[float(x), float(y)] for x, y in bbox]
-
-        text_results.append({
-            "bbox": bbox_py,
-            "text": str(text),
-            "confidence": float(prob)
-        })
-    return text_results
+    # ✅ only text, join all detections
+    extracted_text = " ".join([str(text) for _, text, _ in results])
+    return extracted_text.strip()
 
 
 @app.post("/ocr")
@@ -58,28 +54,27 @@ async def ocr(file: UploadFile = File(...)):
     try:
         contents = await file.read()
 
-        # Detect file type
         if file.filename.lower().endswith(".pdf"):
-            # ✅ Convert PDF to images
+            # ✅ Convert PDF → images
             pages = convert_from_bytes(contents)
 
             # ✅ Run OCR in parallel
             text_results = []
             with ThreadPoolExecutor() as executor:
-                results_list = list(executor.map(run_ocr_on_image, pages))
-                for i, page_results in enumerate(results_list, start=1):
+                page_texts = list(executor.map(run_ocr_on_image, pages))
+                for i, text in enumerate(page_texts, start=1):
                     text_results.append({
                         "page": i,
-                        "results": page_results
+                        "text": text
                     })
 
             return JSONResponse(content={"pdf_results": text_results})
 
         else:
-            # ✅ Normal image case
+            # ✅ Single image case
             image = Image.open(io.BytesIO(contents))
-            text_results = run_ocr_on_image(image)
-            return JSONResponse(content={"results": text_results})
+            text = run_ocr_on_image(image)
+            return JSONResponse(content={"text": text})
 
     except Exception as e:
         return JSONResponse(content={"error": str(e)}, status_code=500)