.

webapp/rescan.py

@@ -81,7 +81,7 @@ def _safe_extract_zip(data: bytes, dest: Path) -> int:
     return count
 
 
-def _run_rescan(job_id: str, staging: Path, prefix: str, force: bool):
+def _run_rescan(job_id: str, staging: Path, prefix: str, force: bool, seed_keywords: str = ""):
     cfg = _cfg()
     try:
         if not cfg["conn"]:
@@ -114,6 +114,7 @@ def _run_rescan(job_id: str, staging: Path, prefix: str, force: bool):
             force=force,
             purge_missing=False,
             remote_root=cfg["remote_root"],
+            seed_keywords=seed_keywords,
         )
         files = []
         if isinstance(summary, dict):
@@ -208,12 +209,14 @@ async def rescan(
     file: list[UploadFile] = File(..., description="ZIP, one help file, or several files from a folder"),
     prefix: str = Form("RIP"),
     force: str = Form("false"),
+    keywords: str = Form(""),
     x_rescan_token: Optional[str] = Header(None, alias="X-Rescan-Token"),
     token: Optional[str] = Form(None),
 ):
     """
     Upload a .zip, one .docx/.html/.pdf/.txt/.doc, or several such files
     (folder pick). Process on server, write sections to OUTPUT_DIR and Postgres.
+    Optional form field ``keywords``: comma-separated terms prepended to every section.
     """
     _check_token(x_rescan_token or token)
 
@@ -232,6 +235,7 @@ async def rescan(
 
     job_id = uuid.uuid4().hex[:12]
     force_flag = str(force).lower() in ("1", "true", "yes", "on")
+    seed_keywords = (keywords or "").strip()
     _set_job(
         job_id,
         status="queued",
@@ -241,7 +245,7 @@ async def rescan(
         force=force_flag,
         created_at=datetime.now(timezone.utc).isoformat(),
     )
-    background_tasks.add_task(_run_rescan, job_id, staging, prefix, force_flag)
+    background_tasks.add_task(_run_rescan, job_id, staging, prefix, force_flag, seed_keywords)
     return JSONResponse(
         {
             "ok": True,