fix: scrapers merge into existing YAML instead of overwriting

scripts/scraper/base_scraper.py

@@ -8,6 +8,7 @@ import urllib.request
 import urllib.error
 from abc import ABC, abstractmethod
 from dataclasses import dataclass, field
+from pathlib import Path
 
 
 @dataclass
@@ -231,7 +232,28 @@ def scraper_cli(scraper_class: type, description: str = "Scrape BIOS requirement
                 if req.zipped_file:
                     entry["zipped_file"] = req.zipped_file
                 config["systems"][sys_id]["files"].append(entry)
-        with open(args.output, "w") as f:
+        # Merge into existing YAML: preserve fields the scraper doesn't generate
+        # (data_directories, case_insensitive_fs, manually added metadata).
+        # The scraper replaces systems + files; everything else is preserved.
+        output_path = Path(args.output)
+        if output_path.exists():
+            with open(output_path) as f:
+                existing = yaml.safe_load(f) or {}
+            # Preserve existing keys not generated by the scraper.
+            # Only keys present in the NEW config are considered scraper-generated.
+            # Everything else in the existing file is preserved.
+            for key, val in existing.items():
+                if key not in config:
+                    config[key] = val
+            # Preserve per-system fields not generated by the scraper
+            # (data_directories, native_id from manual additions, etc.)
+            existing_systems = existing.get("systems", {})
+            for sys_id, sys_data in config.get("systems", {}).items():
+                old_sys = existing_systems.get(sys_id, {})
+                for field in ("data_directories",):
+                    if field in old_sys and field not in sys_data:
+                        sys_data[field] = old_sys[field]
+        with open(output_path, "w") as f:
             yaml.dump(config, f, default_flow_style=False, sort_keys=False)
         print(f"Written {len(reqs)} entries to {args.output}")
         return