Give each help file unique sequential extraction codes, show the count, and allow wiping all extractions for a file while testing.

Co-authored-by: Cursor <cursoragent@cursor.com>

webapp/rescan.py

@@ -102,8 +102,11 @@ def _run_rescan(job_id: str, staging: Path, prefix: str, force: bool):
         # Import here so webapp boots even if processor deps fail later
         from help_processor import process_directory
 
-        process_directory(
-            input_dir=staging,
+        input_dir = staging / "input"
+        if not input_dir.is_dir():
+            input_dir = staging
+        summary = process_directory(
+            input_dir=input_dir,
             output_dir=out,
             conn_str=cfg["conn"],
             api_key=cfg["api_key"],
@@ -112,12 +115,20 @@ def _run_rescan(job_id: str, staging: Path, prefix: str, force: bool):
             purge_missing=False,
             remote_root=cfg["remote_root"],
         )
+        files = []
+        if isinstance(summary, dict):
+            files = summary.get("files") or []
+            sections = summary.get("sections")
+        else:
+            sections = summary
         _set_job(
             job_id,
             status="done",
             message="ok",
             output_dir=str(out),
             prefix=prefix,
+            files=files,
+            sections=sections,
         )
     except Exception as e:
         _set_job(job_id, status="error", message=str(e))