From 168052ed0c3e600a1e62665e2e7a7c33f7f54623 Mon Sep 17 00:00:00 2001
From: nmemmert <67961292+nmemmert@users.noreply.github.com>
Date: Fri, 7 Nov 2025 10:03:47 -0500
Subject: [PATCH] Fix: Validate archive files exist when loading jobs on
 startup

---
 pdf_renamer.py | 30 ++++++++++++++++++++++++++++--
 1 file changed, 28 insertions(+), 2 deletions(-)

diff --git a/pdf_renamer.py b/pdf_renamer.py
index bd33f40..322e7de 100644
--- a/pdf_renamer.py
+++ b/pdf_renamer.py
@@ -92,8 +92,34 @@ def load_jobs():
         if os.path.exists(JOBS_FILE):
             with open(JOBS_FILE, 'r') as f:
                 loaded_jobs = json.load(f)
-                job_queue = loaded_jobs
-                print(f"DEBUG: Loaded {len(job_queue)} jobs from persistent storage")
+                
+                # Only load jobs where archive files still exist
+                valid_jobs = {}
+                skipped_count = 0
+                
+                for job_id, job_data in loaded_jobs.items():
+                    # Only keep completed jobs with existing archive files
+                    if job_data.get('status') == 'completed':
+                        zip_path = job_data.get('zip_path', '')
+                        if zip_path and os.path.exists(zip_path):
+                            valid_jobs[job_id] = job_data
+                        else:
+                            skipped_count += 1
+                            print(f"DEBUG: Skipping job {job_id} - archive file missing: {zip_path}")
+                    # Skip jobs that were in progress (they failed during restart)
+                    elif job_data.get('status') in ['pending', 'processing']:
+                        skipped_count += 1
+                        print(f"DEBUG: Skipping incomplete job {job_id} - status: {job_data.get('status')}")
+                    # Keep error state jobs for visibility
+                    elif job_data.get('status') == 'error':
+                        valid_jobs[job_id] = job_data
+                
+                job_queue = valid_jobs
+                print(f"DEBUG: Loaded {len(job_queue)} jobs from persistent storage ({skipped_count} skipped)")
+                
+                # Save the cleaned up job list back to disk
+                if skipped_count > 0:
+                    save_jobs()
         else:
             print("DEBUG: No existing jobs file found, starting fresh")
     except Exception as e: