Create a dataframe to store the zipcode & price

fp2_11.py

@@ -6,9 +6,14 @@ import json
 import time
 import os
 from concurrent.futures import TimeoutError
+import pandas as pd
 
 # os.environ["GOOGLE_APPLICATION_CREDENTIALS"] = "C:/Users/Karen/AppData/Roaming/gcloud/application_default_credentials.json"
 
+# Create an empty dataframe to store the prices
+df = pd.DataFrame(columns=['Zip Code', 'Price'])
+print(f"Created a dataframe: {df}")
+
 # Configure the Selenium driver options
 options = Options()
 options.add_argument('--disable-gpu')
@@ -121,17 +126,27 @@ def callback(message):
         price = next_data['offers']['price']
         # Print the price
         print(price)
+
+        # Append the zip code and price to the dataframe
+        #append_to_df(zip_code, price)
+        new_row = pd.DataFrame({'Zip Code': [zip_code], 'Price': [price]})
+        return pd.concat([df, new_row], ignore_index=True)
+        #print(f"Recorded the zip code and price: {zip_code},{price}")
+        #print(f"Updated a dataframe: {df}")
+
         subscriber.acknowledge(ack_ids = [message.ack_id], subscription = subscription_path)  # acknowledge receipt of message
     except Exception as e:
         print(f"Caught exception: {e}")
         print(f"Price not found for zip code {zip_code}")
         price = 'Price not found'
+        return df
     
 # Start receiving messages (sync version)
 while True: 
     pull_request = pubsub_v1.types.PullRequest(max_messages = 1, subscription = subscription_path)
     pull_response = subscriber.pull(request = pull_request)
-    callback(pull_response.received_messages[0])
+    df = callback(pull_response.received_messages[0])
+    print(f"Updated a dataframe: {df}")
 
 # Start receiving messages (async version)
 #flow_control = pubsub_v1.types.FlowControl(max_messages = 1)