Adjust audio status timing before waveform is ready

.gitignore

@@ -1,3 +1,12 @@
+.venv/
+.npm-cache/
 __pycache__/
-*.py[cod]
 *.pyo
+*.pyd
+*.py[cod]
+*.log
+debug.log
+sample.wav
+mock_server.py
+rumik-ai-5c44dbeec1e3.json
+*.pyc

app.py

@@ -2,6 +2,7 @@ import datetime as dt
 import json
 import os
 import tempfile
+import time
 from pathlib import Path
 from typing import Any, Dict, Iterator, List, Optional, Tuple
 import uuid
@@ -17,6 +18,7 @@ import requests
 from pymongo import MongoClient
 from pymongo.collection import Collection
 from pymongo.errors import PyMongoError
+from gradio_client import utils as gradio_client_utils
 
 
 MONGODB_URI = os.environ.get("MONGODB_URI")
@@ -26,6 +28,29 @@ USERS_COLLECTION_NAME = os.environ.get("MONGODB_USERS_COLLECTION", "users")
 DEFAULT_LIMIT = int(os.environ.get("CALL_DASHBOARD_LIMIT", "50"))
 SIGNED_URL_EXPIRY_MINUTES = int(os.environ.get("SIGNED_URL_EXPIRY_MINUTES", "60"))
 FALLBACK_BUCKET_NAME = os.environ.get("GCS_BUCKET_NAME")
+AUDIO_READY_DELAY_SECONDS = float(os.environ.get("AUDIO_READY_DELAY_SECONDS", "3.5"))
+
+
+def _patch_gradio_schema_utils() -> None:
+    """Work around gradio_client <1.4 boolean handling for additionalProperties."""
+    original_get_type = gradio_client_utils.get_type
+    original_json_to_python = gradio_client_utils._json_schema_to_python_type
+
+    def _safe_get_type(schema: Any) -> Any:
+        if isinstance(schema, bool):
+            return "object" if schema else "never"
+        return original_get_type(schema)
+
+    def _safe_json_schema_to_python_type(schema: Any, defs: Any) -> Any:
+        if isinstance(schema, bool):
+            return "Dict[str, Any]" if schema else "Never"
+        return original_json_to_python(schema, defs)
+
+    gradio_client_utils.get_type = _safe_get_type
+    gradio_client_utils._json_schema_to_python_type = _safe_json_schema_to_python_type
+
+
+_patch_gradio_schema_utils()
 
 
 if not MONGODB_URI:
@@ -475,11 +500,13 @@ def on_select(
         selection_tracker,
     )
 
+    loaded_from_cache = False
     if bucket_name and blob_name and signed_url:
         cached_path = _AUDIO_CACHE.get(call_id)
         if cached_path and Path(cached_path).exists():
             audio_path = cached_path
             audio_status = "Ready (loaded from cache) — press Play to listen."
+            loaded_from_cache = True
         else:
             try:
                 audio_path = _download_audio(call_id, signed_url, blob_name)
@@ -496,18 +523,43 @@ def on_select(
     else:
         audio_status = "Recording unavailable"
 
-    summary_md = "".join(_build_summary_lines(call, topics, audio_status, signed_url))
     audio_value = gr.update(value=audio_path if audio_path else None, autoplay=False)
-    status_html = (
-        "▶️ Audio ready — press Play to listen."
-        if audio_path
-        else "⚠️ Audio unavailable"
-    )
+    if audio_path:
+        if loaded_from_cache:
+            playback_intermediate = "▶️ Audio ready — press Play to listen."
+            summary_intermediate = "".join(_build_summary_lines(call, topics, audio_status, signed_url))
+        else:
+            playback_intermediate = "⏳ Audio downloaded — finishing waveform analysis…"
+            summary_intermediate = "".join(
+                _build_summary_lines(
+                    call,
+                    topics,
+                    "Downloaded — processing waveform",
+                    signed_url,
+                )
+            )
+    else:
+        playback_intermediate = audio_status if "Error" in audio_status else "⚠️ Audio unavailable"
+        summary_intermediate = "".join(_build_summary_lines(call, topics, audio_status, signed_url))
 
     if selection_tracker.token != selection_token:
         return
 
-    yield audio_value, summary_md, transcription_text, gr.update(value=status_html), selection_tracker
+    yield audio_value, summary_intermediate, transcription_text, gr.update(value=playback_intermediate), selection_tracker
+
+    if not audio_path or loaded_from_cache:
+        return
+
+    if selection_tracker.token != selection_token:
+        return
+
+    time.sleep(max(0.0, AUDIO_READY_DELAY_SECONDS))
+
+    if selection_tracker.token != selection_token:
+        return
+
+    summary_final = "".join(_build_summary_lines(call, topics, audio_status, signed_url))
+    yield gr.update(), summary_final, gr.update(), gr.update(value="▶️ Audio ready — press Play to listen."), selection_tracker
 
 
 with gr.Blocks(title="Call Recording Dashboard") as demo:
@@ -590,6 +642,8 @@ with gr.Blocks(title="Call Recording Dashboard") as demo:
         outputs=[audio_player, call_details, transcription_box, playback_status, selection_tracker],
     )
 
+demo.show_api = False
+
 
 if __name__ == "__main__":
     demo.launch(server_name="0.0.0.0", server_port=int(os.environ.get("PORT", "7860")))