new machine learning api working

machine-learning-client/main.py

@@ -6,6 +6,7 @@
 import json
 from dotenv import load_dotenv
 from bson import ObjectId
+from mindee import Client, PredictResponse, product # To access mindee OCR API
 
 import logging
 
@@ -58,31 +59,37 @@ def pretdict_endpoint():
         return jsonify({'error': 'Object_ID not found in request data'}), 400
 
     Object_ID = ObjectId(request_data['Object_ID']) 
-    logger.debug('OBJECT_ID MESSAGE:', Object_ID)
+    logging.debug('OBJECT_ID MESSAGE:', Object_ID)
     #image = db.receipts.find_one({"_id": Object_ID})['image']
 
     # Here, you would add the code to perform OCR on the image
     # For now, let's assume you have a function called perform_ocr that does this
 
     # Uncomment next line to perform OCR
     data = perform_ocr(Object_ID)
+    logging.debug("data after ocr: %s", data) # debug
+    print("data after ocr: %s", data)
     #data = json.load(open("response1.json", "r"))
 
     # Connect to your collection (replace 'mycollection' with your collection name)
     collection = db['receipts']
 
     # Prepare the data to be inserted into the database
-    receipt = data['receipts'][0]
+    # line_items = data['document']['inference']['pages'][0]['prediction']['line_items']
+    receipt = data['document']['inference']['pages'][0]['prediction']
+    logging.debug('OCR Json Keys:', data['document']['inference']['pages'][0]['prediction'].keys()) # debug
+    print('OCR Json Keys:', data['document']['inference']['pages'][0]['prediction'].keys())
     receipt_data = {
-        'receipt_name': receipt['merchant_name'],
-        'currency': receipt['currency'],
-        'items': [{'description': item['description'], 'amount': item['amount']} for item in receipt['items']],
-        'total': receipt['total'],
-        'tax': receipt['tax'],
-        'tip': receipt['tip'],
-        'subtotal': receipt['subtotal'],
+        'receipt_name': receipt['supplier_name']['raw_value'],
+        'currency': receipt['locale']['currency'],
+        'items': [{'description': item['description'], 'amount': item['total_amount'], 'quantity': item['quantity']} for item in receipt['line_items']],
+        'total': receipt['total_amount']['value'],
+        'tax': receipt['total_tax']['value'],
+        'tip': receipt['tip']['value'],
+        'subtotal': receipt['total_net']['value'],
     }
-    logger.debug(receipt_data)
+    logging.debug(receipt_data)
+    print("receipt_data: %s", receipt_data)
 
     # Update the document with given ObjectId
     collection.update_one({'_id': Object_ID}, {'$set': receipt_data})
@@ -92,32 +99,80 @@ def pretdict_endpoint():
     return jsonify({'_id': str(inserted_id)})
 
 def perform_ocr(Object_ID):
-    logger.debug("starting perform_ocr function...") # debug
-    url = "https://ocr.asprise.com/api/v1/receipt"
+    logging.debug("starting perform_ocr function with mindee api...") # debug
+    url = "https://api.mindee.net/v1/products/mindee/expense_receipts/v5/predict"
+    api_key = os.getenv("OCR_API_KEY")  # Get the API key from environment variable
+
+    # Fetch the image data from the database
     image_data = db.receipts.find_one({"_id": Object_ID})['image']
     file_path = f"receipt_{Object_ID}.jpg"  # Set the file path to save the image
-    logger.debug("file path: %s", file_path) # debug
+    logging.debug("file path: %s", file_path) # debug
+
+    # Save the image data to a file
     with open(file_path, "wb") as f:
         f.write(image_data)
-    file_path = file_path.replace('\x00', '')  # Remove any null bytes from the file path
-
-
-    # Get response (can only do this a couple times with the test API key)
-    res = requests.post(url, 
-                        data = {
-                            'api_key': 'TEST',
-                            'recognizer': 'auto',
-                            'ref_no': 'ocr_python_123'
-                        },
-
-                        files = {
-                            'file': open(file_path, "rb")
-                        })
-
-    with open("response.json", "w") as f:
-        f.write(res.text)
 
-    return res.json()
+    # Parse the file with the Mindee API
+    with open(file_path, "rb") as myfile:
+        files = {"document": myfile}
+        headers = {"Authorization": "Token " + api_key}
+        response = requests.post(url, files=files, headers=headers)
+        print("Type of response: ", type(response))
+        print("response.text: %s", response.text)
+        return response.json()
+
+# def perform_ocr(Object_ID):
+#     logging.debug("starting perform_ocr function with mindee api...") # debug
+#     # Init a new client
+#     api_key = os.getenv("OCR_API_KEY")  # Get the API key from environment variable
+#     mindee_client = Client(api_key)
+
+#     # Fetch the image data from the database
+#     image_data = db.receipts.find_one({"_id": Object_ID})['image']
+#     file_path = f"receipt_{Object_ID}.jpg"  # Set the file path to save the image
+#     logging.debug("file path: %s", file_path) # debug
+
+#     # Save the image data to a file
+#     with open(file_path, "wb") as f:
+#         f.write(image_data)
+
+#     # Load the file
+#     input_doc = mindee_client.source_from_path(file_path)
+
+#     # Parse the file with the Mindee API
+#     result: PredictResponse = mindee_client.parse(product.ReceiptV5, input_doc)
+#     logging.debug("result: %s", result.document) # debug
+
+#     # Return the result as a JSON response
+#     return jsonify(result.document)
+
+# def perform_ocr_old_version(Object_ID):
+#     logging.debug("starting perform_ocr function...") # debug
+#     url = "https://ocr.asprise.com/api/v1/receipt"
+#     image_data = db.receipts.find_one({"_id": Object_ID})['image']
+#     file_path = f"receipt_{Object_ID}.jpg"  # Set the file path to save the image
+#     logging.debug("file path: %s", file_path) # debug
+#     with open(file_path, "wb") as f:
+#         f.write(image_data)
+#     file_path = file_path.replace('\x00', '')  # Remove any null bytes from the file path
+
+
+#     # Get response (can only do this a couple times with the test API key)
+#     res = requests.post(url, 
+#                         data = {
+#                             'api_key': 'TEST',
+#                             'recognizer': 'auto',
+#                             'ref_no': 'ocr_python_123'
+#                         },
+
+#                         files = {
+#                             'file': open(file_path, "rb")
+#                         })
+
+#     with open("response.json", "w") as f:
+#         f.write(res.text)
+
+#     return res.json()
 
 if __name__ == '__main__':
     app.run(host='0.0.0.0', port=5002)  # Run the app

machine-learning-client/requirements.txt

@@ -3,4 +3,5 @@ pylint
 black
 requests
 Flask
-python-dotenv
+python-dotenv
+mindee