fix size accumulation

local-app/python-tools/cross-organization/assess_check_ecr.py

@@ -5,10 +5,11 @@
 from collections import Counter, defaultdict
 
 # --- VERSIONING ---
-__version__ = "1.3.2"
+__version__ = "1.3.3"
 
 def find_latest_file(pattern):
-    files = glob.glob(pattern)
+    """Locates the most recent ECR audit JSON file."""
+    files = glob.glob("audit_results.check_ecr.*.json")
     return max(files, key=os.path.getctime) if files else None
 
 def get_days_ago(iso_str):
@@ -29,7 +30,7 @@ def bucket_age(days, counters):
     else: counters['<30'] += 1
 
 def main():
-    parser = argparse.ArgumentParser(description="ECR Full Spectrum Assessor - v1.3.2")
+    parser = argparse.ArgumentParser(description="ECR Full Spectrum Assessor - v1.3.3")
     parser.add_argument("--input", help="JSON audit file")
     args = parser.parse_args()
 
@@ -54,7 +55,6 @@ def main():
         "org_vulns": Counter(), "scanned_imgs": 0
     }
 
-    # Tracking for duplicates using a Counter for digests
     digest_tracker = Counter()
 
     for idx, account in enumerate(data, 1):
@@ -68,64 +68,63 @@ def main():
             stats["mut"][repo_mut] += 1
             if val.get("has_lifecycle") == "False": stats["no_lc"] += 1
 
-            repo_size = val.get("repo_size_bytes", 0)
-            stats["total_bytes"] += repo_size
-            stats["region_bytes"][region] += repo_size
+            # Tracking per-repo size for the table display
+            repo_size_accum = 0
 
             repo_vulns = Counter()
             images = val.get("images", [])
             stats["total_images"] += len(images)
 
             for img in images:
-                # Security
-                if isinstance(img, list) and len(img) > 0:
-                    img = img[0]
+                if isinstance(img, list) and len(img) > 0: img = img[0]
 
-                # ADD: Record digest for duplicate checking
+                # RESTORED: Storage Aggregation logic
+                img_size = img.get("size_bytes", 0)
+                stats["total_bytes"] += img_size
+                stats["region_bytes"][region] += img_size
+                repo_size_accum += img_size
+
+                # Digest for duplicate checking
                 digest = img.get("image_digest")
-                if digest:
-                    digest_tracker[digest] += 1
+                if digest: digest_tracker[digest] += 1
 
+                # Security & Aging
                 counts = img.get("severity_counts", {})
                 if counts:
                     stats["scanned_imgs"] += 1
                     for sev, count in counts.items():
                         repo_vulns[sev] += count
                         stats["org_vulns"][sev] += count
-                # Aging
+
                 p_days = get_days_ago(img.get("pushed_at"))
                 if p_days is not None:
                     bucket_age(p_days, stats["push_ages"])
                     stats["total_push_days"] += p_days
                     stats["push_count"] += 1
                 bucket_age(get_days_ago(img.get("last_pulled_at")), stats["pull_ages"])
 
-            print(f"{idx:<4} | {account['account_id']:<15} | {region:<12} | {val['repo_name']:<40} | {repo_size/(1024**3):<10.2f} | {repo_mut:<11} | {'YES' if val.get('has_lifecycle')=='True' else 'NO':<7} | {repo_vulns['CRITICAL']:<8} | {repo_vulns['HIGH']:<8}")
+            print(f"{idx:<4} | {account['account_id']:<15} | {region:<12} | {val['repo_name']:<40} | {repo_size_accum/(1024**3):<10.2f} | {repo_mut:<11} | {'YES' if val.get('has_lifecycle')=='True' else 'NO':<7} | {repo_vulns['CRITICAL']:<8} | {repo_vulns['HIGH']:<8}")
 
-    # Footers
+    # Aggregated Summary Calculations
+    total_gb = stats["total_bytes"] / (1024**3)
     avg_img_mb = (stats["total_bytes"] / stats["total_images"]) / (1024**2) if stats["total_images"] > 0 else 0
     avg_push_age = stats["total_push_days"] / stats["push_count"] if stats["push_count"] > 0 else 0
 
     print("-" * report_width)
     print(f"ORGANIZATION ECR SUMMARY\n")
     print(f"  --- Config & Storage ---")
     print(f"  Repos: {stats['total_repos']} ({stats['mut']['IMMUTABLE']} Immutable / {stats['mut']['MUTABLE']} Mutable) | Missing Lifecycle: {stats['no_lc']}")
-    print(f"  Total Data: {stats['total_bytes']/(1024**3):.2f} GB | Average Image Size: {avg_img_mb:.2f} MB\n")
+    print(f"  Total Data: {total_gb:.2f} GB | Average Image Size: {avg_img_mb:.2f} MB\n")
 
     print(f"  --- Security & Vulnerabilities ---")
     print(f"  Scan Coverage: {stats['scanned_imgs']} of {stats['total_images']} images scanned")
     print(f"  Org Totals: " + " | ".join([f"{s}: {stats['org_vulns'][s]:,}" for s in ["CRITICAL", "HIGH", "MEDIUM", "LOW"]]))
 
-    # ADD: Duplicate Image Audit Summary
+    # Duplicate Image Audit
     duplicates = {d: c for d, c in digest_tracker.items() if c > 1}
     print(f"\n  --- Duplicate Image Audit ---")
     print(f"  Total Unique Digests: {len(digest_tracker)}")
     print(f"  Total Duplicated Image Instances: {sum(duplicates.values())} (across {len(duplicates)} unique digests)")
-    if duplicates:
-        print(f"    {'Digest':<75} | {'Instance Count'}")
-        print(f"    {'-'*75} | {'-'*14}")
-        for d, c in sorted(duplicates.items(), key=lambda x: x[1], reverse=True):
-            print(f"    {d:<75} | {c}")
 
     print(f"\n  --- Age Distribution (Days since Push vs Pull) ---")
     print(f"  Average Image Age: {avg_push_age:.1f} days")