Merge pull request #42 from ONS-Innovation/secret_scanning_repo_url

KEH-1060 | Add Repository URL to Secret Scanning Data

Author: TotalDwarf03

data_logger/src/main.py

@@ -899,8 +899,9 @@ def get_secret_scanning_data(logger: wrapped_logging, rest: github_api_toolkit.g
 
             formatted_alert = {
                 "repository": alert["repository"]["name"],
+                "repository_url": alert["repository"]["html_url"],
                 "creation_date": alert["created_at"],
-                "url": alert["html_url"]
+                "alert_url": alert["html_url"],
             }
 
             secret_scanning_data.append(formatted_alert)

docs/data_logger/secret_scanning.md

@@ -12,8 +12,9 @@ This dataset includes all open alerts within the threshold, regardless of whethe
 [    
     {
         "repository": "{repo}",
+        "repository_url": "https://github.com/{org}/{repo}",
         "creation_date": "2024-12-11T23:54:00Z",
-        "url": "https://github.com/{org}/{repo}/security/secret-scanning/{alert_id}"
+        "alert_url": "https://github.com/{org}/{repo}/security/secret-scanning/{alert_id}"
     },
 ]
 ```

src/secret_scanning/collection.py

@@ -32,10 +32,14 @@ def load_secret_scanning(_s3, bucket: str) -> pd.DataFrame | None:
     # Rename the columns to be more readable
     df_secret_scanning.columns = [
         "Repository",
+        "Repository URL",
         "Creation Date",
         "URL",
     ]
 
+    # Remove repository URL since it isn't used (Database requirement only)
+    df_secret_scanning = df_secret_scanning.drop(columns=["Repository URL"])
+
     # Add Alert Age (Days) to the DataFrame
     df_secret_scanning["Alert Age (Days)"] = datetime.now() - pd.to_datetime(df_secret_scanning["Creation Date"]).dt.tz_localize(None)
     df_secret_scanning["Alert Age (Days)"] = df_secret_scanning["Alert Age (Days)"].dt.days