include archived

sync.py

@@ -441,16 +441,47 @@ def get_oldest_unsynced_projects(db, fraction: float = 0.2) -> List[int]:
         return []
 
 
+import json
+
+
+def dump_cases_to_json(projects: List[dict], client: FilevineClient, output_dir: str = "sample-cases", limit: int = 10) -> None:
+    """Fetch project details and dump them as individual JSON files instead of Firestore.
+
+    Args:
+        projects: List of project data dictionaries from Filevine
+        client: FilevineClient instance
+        output_dir: Directory to write JSON files to
+        limit: Maximum number of cases to dump
+    """
+    os.makedirs(output_dir, exist_ok=True)
+    projects = projects[:limit]
+    print(f"[DUMP] Processing {len(projects)} projects to {output_dir}/")
+
+    detailed_rows = process_projects_parallel(projects, client, max_workers=10)
+    written = 0
+    for row in detailed_rows:
+        if row.get('ProjectId'):
+            row['is_archived'] = (row.get('phase_name') == 'Archived')
+            filepath = os.path.join(output_dir, f"{row['ProjectId']}.json")
+            with open(filepath, 'w', encoding='utf-8') as f:
+                json.dump(row, f, indent=2, default=str)
+            written += 1
+            print(f"[DUMP] Wrote {filepath}")
+
+    print(f"[DUMP] Complete - {written} cases written to {output_dir}/")
+
+
 def main():
     """Main function to fetch and sync projects"""
     import argparse
 
     parser = argparse.ArgumentParser(description='Sync Filevine projects to Firestore')
-    parser.add_argument('--mode', choices=['full', 'last_n', 'oldest_percent', 'hybrid', 'single'],
-                        default='hybrid', help='Sync mode: full=all projects, last_n=recently active, oldest_percent=oldest by last_synced_at, hybrid=last_n+oldest_percent, single=one project')
+    parser.add_argument('--mode', choices=['full', 'last_n', 'oldest_percent', 'hybrid', 'single', 'dump_json'],
+                        default='hybrid', help='Sync mode: full=all projects, last_n=recently active, oldest_percent=oldest by last_synced_at, hybrid=last_n+oldest_percent, single=one project, dump_json=export to JSON files')
     parser.add_argument('--days', type=int, default=14, help='Number of days for last_n mode (default: 14)')
     parser.add_argument('--percent', type=float, default=20.0, help='Percentage for oldest_percent mode (default: 20)')
     parser.add_argument('--project-id', type=int, help='Project ID for single mode (required when mode=single)')
+    parser.add_argument('--limit', type=int, default=10, help='Number of cases to dump (default: 10)')
     args = parser.parse_args()
 
     if args.mode == 'single' and not args.project_id:
@@ -460,11 +491,14 @@ def main():
     try:
         client = FilevineClient()
         client.get_bearer_token()
-        from app import db
 
         recent_successes = 0
         oldest_successes = 0
         total_failures = 0
+        documents = []
+
+        if args.mode != 'dump_json':
+            from app import db
 
         if args.mode == 'full':
             print("[MODE] Full sync - fetching all projects")
@@ -561,7 +595,13 @@ def main():
             total_failures = len(detailed_rows) - len(project_ids_synced)
             record_sync_stats(db, recent_successes, oldest_successes, total_failures)
 
-        print(f"[SYNC] Complete - {len(documents)} projects saved to Firestore")
+        elif args.mode == 'dump_json':
+            print(f"[MODE] Dump JSON - fetching {args.limit} cases")
+            projects = client.list_all_projects()
+            dump_cases_to_json(projects, client, limit=args.limit)
+
+        if args.mode != 'dump_json':
+            print(f"[SYNC] Complete - {len(documents)} projects saved to Firestore")
 
     except Exception as e:
         print(f"Error during sync: {e}")