clean user coolection file

scripts/user_collection.py

@@ -7,55 +7,32 @@ import pandas as pd
 #import datetime
 import sys
 from datetime import datetime
+import matplotlib.pyplot as plt
 
-# Initialize Firebase Admin SDK
 cred = credentials.Certificate(os.getcwd() + '/keys/hpos-prod-firebase-adminsdk.json')  # Replace with your own service account key path
 firebase_admin.initialize_app(cred)
 
-# Get a reference to the Firestore database
 db = firestore.client()
 
-# Specify the collections
 patient_collection = db.collection("patientData")
-test_collection = db.collection("testData")
-start_date = sys.argv[1] # '2023-01-12' #input("Please enter the start date (yyyy-mm-dd): ")
-end_date = sys.argv[2] #'2023-07-16' #input("Please enter the end date (yyyy-mm-dd): ")
+# test_collection = db.collection("testData")
+start_date = sys.argv[1]
+end_date = sys.argv[2]
 
 query = patient_collection.where(filter=FieldFilter("createdAt", ">=", start_date)).where(filter=FieldFilter("createdAt", "<", end_date))
 patient_docs = query.stream()
 
-# Prepare data to store in CSV
 data = []
 for patient_doc in patient_docs:
     patient_data = patient_doc.to_dict()
-    # patient_id = patient_data["_id"]
-
-    # # Query the document from testData collection based on the common _id
-    # test_docs = test_collection.where("_id", "==", patient_id).stream()
-
-    # for test_doc in test_docs:
-    #     print(test_doc)
-    #     test_data = test_doc.to_dict()
-
-    #     # Combine the data from both collections into a single dictionary
-    #     combined_data = {**patient_data, **test_data}
-
-    #     # Fill empty fields in test_data with corresponding values from patient_data
-    #     for key, value in combined_data.items():
-    #         if value == "" and key in patient_data:
-    #             combined_data[key] = patient_data[key]
-
-    #     data.append(combined_data)
     data.append(patient_data)
 
-# Convert the data to a DataFrame
 df = pd.DataFrame(data)
 print(df)
 print(df.size)
 
 print("duplicates", len(df['_id']) - len(df['_id'].drop_duplicates()))
 
-# Save the DataFrame to a CSV file
 output_filename =  f'users_{datetime.today().strftime("%d_%m_%Y_%H_%M")}.xlsx'
 df.to_excel(output_filename, index=False)
 
@@ -65,6 +42,5 @@ df.to_excel(output_path, index=False)
 
 print(f"Data saved to '{output_path}'")
 
-# Close the Firebase Admin SDK
 firebase_admin.delete_app(firebase_admin.get_app())