.

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,

webapp/templates/viewer.html

@@ -276,6 +276,8 @@
   .rescan-filemeta.warn { color: var(--danger); }
   .wipe-box { margin-top: 16px; }
   .wipe-box input[type=text] { width: 100%; font-family: var(--mono); }
+  .rescan-kw-input { width: 100%; }
+  .rescan-hint { margin-bottom: 10px; line-height: 1.45; }
   .src-file { cursor: pointer; }
   .src-file:hover { color: var(--accent); text-decoration: underline; }
 
@@ -361,6 +363,15 @@
            onchange="onScanFolderPicked()">
     <div class="rescan-selection" id="rescan-selection"></div>
     <div class="rescan-filemeta" id="rescan-filemeta"></div>
+    <p class="stats rescan-hint">
+      Ръчни ключови думи (раздели със запетая) — записват се <strong>първи</strong> във всяка секция
+      от това сканиране. Ако файлът вече е сканиран, включи force.
+    </p>
+    <div class="rescan-row">
+      <input type="text" id="rescan-keywords" class="rescan-kw-input"
+             placeholder="напр. лимитни карти, справка, ATR"
+             autocomplete="off">
+    </div>
     <div class="rescan-row">
       <label class="stats"><input type="checkbox" id="rescan-force"> force</label>
       <button type="button" class="btn-primary" id="rescan-btn" onclick="startRescan()">Старт</button>
@@ -486,7 +497,7 @@
       <div class="api-ep"><span class="method post">POST</span><code>/api/keywords/{code}</code>
         <div class="desc">Body JSON: <code>{"keywords":"а, б, в"}</code> — обновява ключови думи.</div></div>
       <div class="api-ep"><span class="method post">POST</span><code>/api/rescan</code>
-        <div class="desc">multipart: <code>file</code> (един или повече: zip/docx/…), <code>prefix</code>, <code>force</code>. Header опционално: <code>X-Rescan-Token</code>.</div></div>
+        <div class="desc">multipart: <code>file</code> (един или повече: zip/docx/…), <code>prefix</code>, <code>force</code>, по избор <code>keywords</code> (запетая; първи във всяка секция). Header опционално: <code>X-Rescan-Token</code>.</div></div>
       <div class="api-ep"><span class="method">GET</span><code>/api/rescan/{job_id}</code>
         <div class="desc">Статус на rescan задача (<code>queued</code> / <code>processing</code> / <code>done</code> / <code>error</code>).</div></div>
       <div class="api-ep"><span class="method">GET</span><code>/api/file-extractions?file=…&amp;prefix=RIP</code>
@@ -866,6 +877,7 @@ async function startRescan() {
   scanFiles.forEach(f => fd.append('file', f));
   fd.append('prefix', PAGE_PREFIX);
   fd.append('force', document.getElementById('rescan-force').checked ? 'true' : 'false');
+  fd.append('keywords', (document.getElementById('rescan-keywords').value || '').trim());
   const token = localStorage.getItem('RESCAN_TOKEN') || '';
   if (token) fd.append('token', token);