Update the script to include more reports

README.md

@@ -22,6 +22,8 @@ Also check
 - Python 3
 - `googleapiclient`
 - `oauth2client`
+- `pyyaml`
+- `argparse`
 
 ## Setup and Installation
 
@@ -40,7 +42,7 @@ It's recommended to use a virtual environment:
 ```sh
 python -m venv venv
 source venv/bin/activate
-pip install google-api-python-client oauth2client
+pip install google-api-python-client oauth2client pyyaml argparse
 ```
 
 3. **Service Account and API Key:**
@@ -49,17 +51,19 @@ pip install google-api-python-client oauth2client
 - Rename the file to `ua-archive.json` and place it in the project directory.
 
 4. **Configure the Script:**
-- Open the script in a text editor.
-- Modify the parameters at the bottom of the script (`api_key`, `view_id`, `dimensions`, `metrics`, `start_date`, `end_date`, `output_file`) as needed.
+- Copy 'settings.yml.defualt' to 'settings.yml' and update 'api_key' with the name of json file generated and 'view_id' with your UA property's view ID.
+- Edit the `reports_config.yml` to add the reports you want to generate.
 
 ## Usage
 
 Run the script with Python:
 
 ```sh
-python ua_archive.py
+python3 analytics_reporter.py --report_id 1 --start 2023-01-01 --end 2023-01-31
 ```
-After successful execution, a CSV file named `UA_report.csv` will be generated in the project directory.
+report_id should be the id in the `reports_config.yml`
+
+After successful execution, a CSV file named `<report_id>_<report_name>_report.csv` will be generated in the project directory.
 
 ## Contributing
 

analytics_reporter.py

@@ -0,0 +1,62 @@
+"""
+Analytics Reporter
+
+Author: Vimal Joseph
+More Information: https://www.zyxware.com/article/6662/backup-universal-analytics-data-python
+
+This script generates reports based on Google Analytics data.
+"""
+import argparse
+import yaml
+from ga_data_fetcher import get_data
+from utils import format_date, write_to_csv
+
+def load_yaml_config(file_path):
+    with open(file_path, 'r') as file:
+        return yaml.safe_load(file)
+
+def generate_report(report_config, start_date, end_date, api_key, view_id, output_file):
+    if not report_config:
+        print("Report configuration not found.")
+        return
+
+    dimensions = report_config['dimensions']
+    metrics = report_config['metrics']
+
+    data = get_data(api_key, view_id, dimensions, metrics, start_date, end_date, format_date)
+
+    if data:
+        write_to_csv(data, output_file)
+        print(f"Successfully saved report to CSV file: {output_file}")
+    else:
+        print("No data received for the report.")
+
+def main():
+    parser = argparse.ArgumentParser(description='Generate Google Analytics reports.')
+    parser.add_argument('-id', '--report_id', type=int, required=True, help='ID of the report to generate')
+    parser.add_argument('-s', '--start', type=str, required=True, help='Start date (YYYY-MM-DD)')
+    parser.add_argument('-e', '--end', type=str, required=True, help='End date (YYYY-MM-DD)')
+    args = parser.parse_args()
+
+    settings = load_yaml_config("settings.yml")
+    report_configs = load_yaml_config(settings['analytics_settings']['reports_config'])
+
+    report_id = args.report_id
+    start_date = args.start
+    end_date = args.end
+    api_key = settings['analytics_settings']['api_key']
+    view_id = settings['analytics_settings']['view_id']
+
+    report_config = next((r for r in report_configs['reports'] if r['id'] == report_id), None)
+    if not report_config:
+        print(f"Report configuration for ID {report_id} not found.")
+        return
+    # Construct the file name
+    report_name = report_config['name'].replace(' ', '-').lower()
+    output_file = f"{report_id}_{report_name}_report.csv"
+    generate_report(report_config, start_date, end_date, api_key, view_id, output_file)
+
+
+if __name__ == "__main__":
+    main()
+

ga_data_fetcher.py

@@ -0,0 +1,58 @@
+"""
+Data Fetcher
+
+Author: Vimal Joseph
+More Information: https://www.zyxware.com/article/6662/backup-universal-analytics-data-python
+
+This script fetch the required data using Google Analytics API.
+"""
+from googleapiclient.discovery import build
+from oauth2client.service_account import ServiceAccountCredentials
+from googleapiclient.errors import HttpError
+
+def get_data(api_key, view_id, dimensions, metrics, start_date, end_date, date_formatter):
+    # Initialize service
+    credentials = ServiceAccountCredentials.from_json_keyfile_name(api_key)
+    service = build('analyticsreporting', 'v4', credentials=credentials)
+
+    # Build request
+    request = service.reports().batchGet(
+        body={
+            'reportRequests': [
+                {
+                    'viewId': view_id,
+                    'dateRanges': [{'startDate': start_date, 'endDate': end_date}],
+                    'dimensions': [{'name': d} for d in dimensions],
+                    'metrics': [{'expression': m} for m in metrics]
+                }
+            ]
+        }
+    )
+
+    try:
+        response = request.execute()
+        report = response.get('reports', [])[0]  # Assuming one report request
+
+        column_header_entries = report['columnHeader']['dimensions'] + \
+                                [entry['name'] for entry in report['columnHeader']['metricHeader']['metricHeaderEntries']]
+
+        rows = report.get('data', {}).get('rows', [])
+
+        formatted_data = []
+        for row in rows:
+            formatted_row = {}
+            dimensions_data = row['dimensions']
+            metrics_data = row['metrics'][0]['values']
+            all_data = dimensions_data + metrics_data
+            for i, header in enumerate(column_header_entries):
+                if header == 'ga:date':
+                    formatted_row[header] = date_formatter(all_data[i])
+                else:
+                    formatted_row[header] = all_data[i]
+            formatted_data.append(formatted_row)
+
+        return formatted_data
+    except HttpError as error:
+        print(f"Error fetching data: {error}")
+        return []
+

reports_config.yml

@@ -0,0 +1,46 @@
+reports:
+  - id: 1
+    name: "User Engagement Over Time"
+    dimensions: ["ga:clientId", "ga:date"]
+    metrics: ["ga:sessions", "ga:sessionDuration", "ga:pageviews"]
+
+  - id: 2
+    name: "User Journey Analysis"
+    dimensions: ["ga:landingPagePath", "ga:exitPagePath", "ga:pagePath", "ga:previousPagePath"]
+    metrics: ["ga:sessions", "ga:pageviews", "ga:entrances", "ga:exits"]
+
+  - id: 3
+    name: "Acquisition Channels and Campaign Performance"
+    dimensions: ["ga:source", "ga:medium", "ga:campaign"]
+    metrics: ["ga:sessions", "ga:users"]
+
+  - id: 4
+    name: "User Segmentation"
+    dimensions: ["ga:userAgeBracket", "ga:userGender", "ga:userType"]
+    metrics: ["ga:users", "ga:sessions", "ga:sessionDuration", "ga:pageviews"]
+
+  - id: 5
+    name: "Conversion Analysis"
+    dimensions: ["ga:goalCompletionLocation", "ga:goalPreviousStep1", "ga:goalPreviousStep2", "ga:goalPreviousStep3"]
+    metrics: ["ga:goalCompletions", "ga:goalConversionRate"]
+
+  - id: 6
+    name: "User Interaction with Content"
+    dimensions: ["ga:pagePath"]
+    metrics: ["ga:pageviews", "ga:uniquePageviews", "ga:avgTimeOnPage"]
+
+  - id: 7
+    name: "Keyword and SEO Analysis"
+    dimensions: ["ga:keyword", "ga:searchQuery"]
+    metrics: ["ga:sessions", "ga:users", "ga:organicSearches"]
+
+  - id: 8
+    name: "Device and Technology Analysis"
+    dimensions: ["ga:deviceCategory", "ga:browser", "ga:operatingSystem"]
+    metrics: ["ga:sessions", "ga:users", "ga:pageviews"]
+
+  - id: 9
+    name: "Custom User Events Tracking"
+    dimensions: ["ga:eventCategory", "ga:eventAction", "ga:eventLabel"]
+    metrics: ["ga:totalEvents", "ga:uniqueEvents"]
+

settings.yml.default

@@ -0,0 +1,5 @@
+analytics_settings:
+  reports_config: "reports_config.yml"
+  api_key: "REPLACE with JSON file name from Google"
+  view_id: "REPLACE with UA View ID"
+

ua_archive.py

@@ -1,3 +1,11 @@
+"""
+Analytics Reporter
+
+Author: Vimal Joseph
+More Information: https://www.zyxware.com/article/6662/backup-universal-analytics-data-python
+
+This script generates reports based on Google Analytics data.
+"""
 import csv
 from googleapiclient.discovery import build
 from oauth2client.service_account import ServiceAccountCredentials

utils.py

@@ -0,0 +1,18 @@
+import csv
+def format_date(date_str):
+    """Convert date from 'YYYYMMDD' to 'YYYY-MM-DD' format."""
+    try:
+        from datetime import datetime
+        return datetime.strptime(date_str, '%Y%m%d').strftime('%Y-%m-%d')
+    except ValueError:
+        return date_str  # Return the original string if it can't be converted
+
+def write_to_csv(data, output_file):
+    if not data:
+        print("No data to write to CSV.")
+        return
+
+    with open(output_file, 'w', newline='') as csvfile:
+        writer = csv.DictWriter(csvfile, fieldnames=data[0].keys())
+        writer.writeheader()
+        writer.writerows(data)