Adding ability to pause transcription

scripts/podx-tools.sh

@@ -85,6 +85,9 @@ fi
 # Optional: explicit KB id to use for all KB operations (bypasses name resolution)
 : "${OPENWEBUI_KB_ID:=}"
 
+# Max seconds to wait for OWUI to finish extracting file content before attach
+: "${OPENWEBUI_WAIT_SECS:=180}"
+
 # Resolve a working OpenWebUI base URL (fallback from host.docker.internal -> localhost)
 _owui_url() {
   # If a host-only override is provided, prefer it unconditionally
@@ -225,6 +228,21 @@ EOF
   done
 }
 
+# --- Redis helpers (for pause/resume switches) -----------------------------
+_redis_cli() {
+  # Prefer dockerized redis if available; otherwise fall back to local redis-cli
+  if command -v docker >/dev/null 2>&1 && docker compose ps redis >/dev/null 2>&1; then
+    docker compose exec -T redis redis-cli "$@"
+  elif command -v redis-cli >/dev/null 2>&1; then
+    redis-cli -h "${REDIS_HOST:-127.0.0.1}" -p "${REDIS_PORT:-6379}" "$@"
+  else
+    echo "redis-cli not found and no docker compose redis service available." >&2
+    return 1
+  fi
+}
+
+_transcribe_key="podx:transcribe:paused"
+
 _help() {
   cat <<EOF
 podx-tools — Meili & OpenWebUI helpers
@@ -252,6 +270,11 @@ OpenWebUI:
   owui-kbs-raw                       # raw JSON from /knowledge/list
   owui-batch-attach "<KB Name>" <glob>  # attach all files matching glob
 
+Transcription control:
+  transcribe-status                 # show whether transcription workers are paused
+  transcribe-pause                  # pause CPU-heavy transcription jobs
+  transcribe-resume                 # resume transcription jobs
+
 Examples:
   ./scripts/podx-tools.sh meili-health
   ./scripts/podx-tools.sh meili-search "grand canyon" 10
@@ -451,6 +474,7 @@ PY
     _require "OPENWEBUI_URL" "$OPENWEBUI_URL"
     _require "OPENWEBUI_API_KEY" "$OPENWEBUI_API_KEY"
 
+    TMP_EXTRACT=""
     # Decide how to send the file (force text/plain for .txt/.md; optionally extract JSON->text)
     upload_flag=("-F" "file=@$file")
     ext="${file##*.}"
@@ -475,6 +499,7 @@ PY
             stem="${base%.*}"
             upload_flag=("-F" "file=@$tmp_txt;type=text/plain;filename=${stem}.txt")
             echo "[owui] extracted text from JSON -> ${stem}.txt"
+            TMP_EXTRACT="$tmp_txt"
           else
             echo "[owui] WARNING: JSON had no extractable text, uploading raw JSON (may be rejected)" >&2
           fi
@@ -507,7 +532,7 @@ PY
     if [ -z "$FILE_ID" ]; then echo "Upload failed (no file id)"; exit 1; fi
 
     # Wait until OWUI finishes processing/extracting text for this file (prevents "content empty" 400)
-    if ! _owui_wait_file "$FILE_ID" 180; then
+    if ! _owui_wait_file "$FILE_ID" "$OPENWEBUI_WAIT_SECS"; then
       echo "[owui] WARNING: timed out waiting for file content; attach may fail if OWUI hasn't extracted text yet" >&2
     fi
 
@@ -537,6 +562,7 @@ PY
     echo "$RESP" | ppjson
 
     rm -f "$tmp_body" "$tmp_code" "$tmp_hdrs"
+    [ -n "${TMP_EXTRACT:-}" ] && rm -f "$TMP_EXTRACT"; unset TMP_EXTRACT
 
     if [ $curl_exit -ne 0 ]; then
       echo "Attach failed: curl exit $curl_exit" >&2; exit $curl_exit
@@ -550,6 +576,10 @@ PY
         : # success
         ;;
       *)
+        if printf '%s' "$RESP" | grep -qi "Duplicate content"; then
+          echo "[owui] duplicate content — already indexed. Treating as success."
+          exit 0
+        fi
         echo "Attach failed (HTTP $http_code)" >&2; exit 1
         ;;
     esac
@@ -564,6 +594,7 @@ PY
     _require "OPENWEBUI_URL" "$OPENWEBUI_URL"
     _require "OPENWEBUI_API_KEY" "$OPENWEBUI_API_KEY"
 
+    TMP_EXTRACT=""
     # Decide how to send the file (force text/plain for .txt/.md; optionally extract JSON->text)
     upload_flag=("-F" "file=@$file")
     ext="${file##*.}"
@@ -588,6 +619,7 @@ PY
             stem="${base%.*}"
             upload_flag=("-F" "file=@$tmp_txt;type=text/plain;filename=${stem}.txt")
             echo "[owui] extracted text from JSON -> ${stem}.txt"
+            TMP_EXTRACT="$tmp_txt"
           else
             echo "[owui] WARNING: JSON had no extractable text, uploading raw JSON (may be rejected)" >&2
           fi
@@ -620,7 +652,7 @@ PY
     if [ -z "$FILE_ID" ]; then echo "Upload failed (no file id)"; exit 1; fi
 
     # Wait until OWUI finishes processing/extracting text for this file (prevents "content empty" 400)
-    if ! _owui_wait_file "$FILE_ID" 180; then
+    if ! _owui_wait_file "$FILE_ID" "$OPENWEBUI_WAIT_SECS"; then
       echo "[owui] WARNING: timed out waiting for file content; attach may fail if OWUI hasn't extracted text yet" >&2
     fi
 
@@ -640,6 +672,7 @@ PY
     echo "$RESP" | ppjson
 
     rm -f "$tmp_body" "$tmp_code" "$tmp_hdrs"
+    [ -n "${TMP_EXTRACT:-}" ] && rm -f "$TMP_EXTRACT"; unset TMP_EXTRACT
 
     if [ $curl_exit -ne 0 ]; then
       echo "Attach failed: curl exit $curl_exit" >&2; exit $curl_exit
@@ -649,7 +682,13 @@ PY
     fi
     case "$http_code" in
       200|201|204) : ;;
-      *) echo "Attach failed (HTTP $http_code)" >&2; exit 1 ;;
+      *)
+        if printf '%s' "$RESP" | grep -qi "Duplicate content"; then
+          echo "[owui] duplicate content — already indexed. Treating as success."
+          exit 0
+        fi
+        echo "Attach failed (HTTP $http_code)" >&2; exit 1
+        ;;
     esac
     ;;
 
@@ -773,6 +812,31 @@ PY
     done
     ;;
 
+  transcribe-status)
+    # Prints 'paused' if the switch key is set to a truthy value, else 'running'
+    val="$(_redis_cli GET "$_transcribe_key" 2>/dev/null || true)"
+    if [ -n "${val:-}" ] && [ "${val}" != "(nil)" ] && [ "${val}" != "0" ]; then
+      echo "paused"
+      exit 0
+    fi
+    echo "running"
+    ;;
+  transcribe-pause)
+    if _redis_cli SET "$_transcribe_key" 1 >/dev/null; then
+      echo "Transcription: paused"
+    else
+      echo "Failed to talk to redis to set pause switch." >&2
+      exit 1
+    fi
+    ;;
+  transcribe-resume)
+    if _redis_cli DEL "$_transcribe_key" >/dev/null; then
+      echo "Transcription: resumed"
+    else
+      echo "Failed to talk to redis to clear pause switch." >&2
+      exit 1
+    fi
+    ;;
   *)
     echo "Unknown command: $cmd" >&2
     _help