Render waveform in a different context (#10)

* add firestore_sdk ad session_manager

* save user chats on firestore

* pass down session_id for a deterministic workflow

* handle conversion of chat object to/fro a dict

* remove references to langchain

* reuse a previously downloaded audiofile if it's processable

* render audiocast metdata on share page

* cleanup

* temp remove audio_enchancement

* sanitize audiocast transcript

* add elevenlabs client

* add __text_to_speech_elevenlabs; cleanup

* use dry in text_to_speech

* only lint on python versions 3.11 and 3.12

* add write permission to deploy job for marocchino/sticky-pull-request-comment

* use eleven_multilingual_v2 model for improved stability, accuracy and quality

* Refactor audiocast page to include waveform visualization

* put waveform viz in an expander

* cleanup

* move download_waveform_video internal to render_waveform

* allow toggling waveform visualizer

* save waveform to gcs

* reshuffle dependencies in requirements.txt

* add pycairo to deps

* add a server app

* add header and process_time middlewares

* make utils a shared folder

* use root level services in app dir

* write an improved init_project workflow

* move streamlit related packages into app folder

* convert app.src.utils into a module

* install setup tools and create a setup file for utils_pkg

* remove streamlit content from chat_utils

* cleanup remnants

* more cleanup

* update package imports and initialization logic for utils_pkg

* hit the backend in for chat workflow in generate_stream_response

* add generate_audiocast endpoint handler

* create a handler for get_audiocast by session_id

* abstract waveform_utils; save wave_form on_generate audiocast

* hit the server for generate_audiocast and get_audiocast

* abstract audiocast inteface definitions

* move audiocast_request to server

* remove unused artifacts

* save and update metadata.info

* Remove audiocast page and add docstrings to session manager methods

* revise github actions workflow files

* add deploy_server.yml

* rename deploy server workflow to reflect server deployment

* upgrade to the latest version of pip

* cleanup

* remove the underscore in the service name

* remove unused env variables

* cleanup

* add dockerfile for deploying the server

* populate a dockerignore file on the fly

* remove unused context from gcloudignore

* localize the .*ignore files

* move the dockerfile to root project pre-deploy step

* specify corret path to streamlit index

* Add debug output for Docker and gcloud ignore files; set no_traffic to false for deployment

* add a gunicorn config file

* rename server to api

* rename all footprints of server to api

* refactor: update workflow names and paths for cloudrun deployment

* bump version to 1.0.1 and update project description

* move chat_request to api utils

* move api only utils into api folder

* rename utils_pkg to shared_utils_pkg

* add pre-commit config

* refactor: clean up sidebar and improve metadata subscription handling

* feat: add Streamlit configuration for browser and client settings

* log the project_id in init_admin_sdk

* remove reference_code from .gitignore

* allow other ui elements to render before loading waveform video

* remove subscription to metadata.info

app/pages/audiocast.py

@@ -32,7 +32,7 @@ async def render_audiocast_page():
             if audiocast["created_at"]:
                 st.markdown(f"> Created: {audiocast["created_at"]}")
 
-            share_url = render_audiocast_handler(session_id, audiocast)
+            share_url, finalize_task = render_audiocast_handler(session_id, audiocast)
 
             share_col, restart_row = st.columns(2, vertical_alignment="center")
             with share_col:
@@ -42,6 +42,7 @@ async def render_audiocast_page():
                 if st.button("Create your Audiocast", use_container_width=True):
                     navigate_to_home()
 
+            await finalize_task
         except Exception as e:
             st.error(f"Error loading audiocast: {str(e)}")
     else:

app/src/utils/metadata_subscription.py

@@ -1,29 +1,34 @@
-from queue import Queue
+import asyncio
 
 import streamlit as st
 
 from shared_utils_pkg.session_manager import SessionManager
 
 
-def subscribe_to_audio_generation(session_id: str):
-    """Subscribe to audio generation metadata"""
-    q = Queue()
+class SubscribeToAudioGeneration:
+    def __init__(self, session_id: str):
+        self.session_id = session_id
+        self.info: str | None = None
 
-    def handler(info: str | None):
-        if info:
-            q.put(info, block=False)
+    def create(self):
+        """Subscribe to audio generation metadata"""
 
-    db = SessionManager(session_id)
-    doc_watch = db.subscribe_to_metadata_info(handler)
+        def __handler(info: str):
+            self.info = info
 
-    with st.empty():
-        while True:
-            try:
-                info = q.get(timeout=2)
-                if not info:
-                    break
-                st.info(info)
-            except Exception:
-                break
+        db = SessionManager(self.session_id)
+        return db.subscribe_to_metadata_info(__handler)
 
-    return doc_watch
+    async def _update_ui(self, read_timeout=10):
+        """Update the UI with the current info attribute value."""
+        timeout = 0
+        container = st.empty()
+
+        with container:
+            while read_timeout > timeout:
+                if self.info:
+                    container.info(self.info)
+                await asyncio.sleep(1)
+                timeout += 1
+
+        container.empty()

app/src/utils/render_audiocast.py

@@ -1,5 +1,4 @@
 import streamlit as st
-
 from src.utils.custom_components import copy_button
 from src.utils.render_audiocast_utils import (
     GenerateAudiocastDict,
@@ -8,15 +7,15 @@
 from src.utils.session_state import reset_session
 
 
-def render_audiocast(session_id: str):
+async def render_audiocast(session_id: str):
     """
     Render the audiocast based on the user's preferences
     - Display current audiocast if available
     """
     st.markdown("#### Your Audiocast")
     current_audiocast: GenerateAudiocastDict = st.session_state.current_audiocast
 
-    share_url = render_audiocast_handler(session_id, current_audiocast)
+    share_url, finalize_task = render_audiocast_handler(session_id, current_audiocast)
 
     share_col, restart_row = st.columns(2, vertical_alignment="center")
 
@@ -28,3 +27,4 @@ def render_audiocast(session_id: str):
         if st.button("Generate New Audiocast", use_container_width=True):
             reset_session()
             st.rerun()
+    await finalize_task

app/src/utils/render_audiocast_utils.py

@@ -1,11 +1,11 @@
+import asyncio
 import re
 from pathlib import Path
 from typing import cast
 
 import httpx
 import streamlit as st
-from src.utils.metadata_subscription import subscribe_to_audio_generation
-from src.utils.render_waveform import render_waveform
+from src.utils.render_waveform import load_waveform_video, render_waveform
 
 from env_var import API_URL, APP_URL
 from shared_utils_pkg.audiocast_utils import GenerateAudioCastRequest, GenerateAudiocastDict
@@ -33,8 +33,6 @@ async def generate_audiocast(
     summary: str,
     content_category: ContentCategory,
 ):
-    doc_watch = subscribe_to_audio_generation(session_id)
-
     audiocast_req = GenerateAudioCastRequest(
         sessionId=session_id,
         summary=summary,
@@ -46,8 +44,6 @@ async def generate_audiocast(
         timeout=None,
     )
     response.raise_for_status()
-    doc_watch.unsubscribe()
-
     return cast(GenerateAudiocastDict, response.json())
 
 
@@ -57,10 +53,11 @@ def render_audiocast_handler(session_id: str, audiocast: GenerateAudiocastDict):
 
     # Voice waveform
     with st.expander("Show Audio Waveform"):
-        try:
-            render_waveform(session_id, audiocast["url"], False)
-        except Exception as e:
-            st.error(f"Error rendering waveform: {str(e)}")
+        waveform_video_path: Path | str = st.session_state.get("waveform_video_path", False)
+        if waveform_video_path:
+            render_waveform(waveform_video_path)
+        else:
+            st.info("Loading waveform video...")
 
     # Transcript
     with st.expander("Show Transcript"):
@@ -75,4 +72,11 @@ def render_audiocast_handler(session_id: str, audiocast: GenerateAudiocastDict):
     share_url = f"{APP_URL}/audiocast?session_id={session_id}"
     st.text_input("Share this audiocast:", share_url)
 
-    return share_url
+    async def finalize():
+        try:
+            load_waveform_video(session_id, audiocast["url"])
+        except Exception as e:
+            st.error(f"Error rendering waveform: {str(e)}")
+
+    finalize_task = asyncio.create_task(finalize())
+    return share_url, finalize_task

app/src/utils/render_waveform.py

@@ -1,4 +1,5 @@
 import os
+from pathlib import Path
 
 import streamlit as st
 from pydub import AudioSegment
@@ -7,8 +8,8 @@
 from shared_utils_pkg.waveform_utils import WaveformUtils
 
 
-def render_waveform(session_id: str, audio_path: str, autoplay=False):
-    """Render waveform visualization from audio file."""
+def load_waveform_video(session_id: str, audio_path: str):
+    """Load waveform visualization from audio file."""
     waveform_utils = WaveformUtils(session_id, audio_path)
     tmp_vid_path = waveform_utils.get_tmp_video_path()
 
@@ -32,15 +33,22 @@ def render_waveform(session_id: str, audio_path: str, autoplay=False):
                 video_path = waveform_utils.generate_waveform_video(tmp_vid_path)
                 waveform_utils.save_waveform_video_to_gcs(str(video_path))
 
-        with open(video_path, "rb") as video_file:
-            video_bytes = video_file.read()
-            st.video(video_bytes, autoplay=autoplay)
-
-        download_waveform_video(str(video_path))
+        if not st.session_state.get("waveform_video_path"):
+            st.session_state.waveform_video_path = video_path
+            st.rerun()
     except Exception as e:
         st.error(f"Error generating visualization: {str(e)}")
 
 
+def render_waveform(video_path: Path | str, autoplay=False):
+    """Render waveform visualization from path"""
+    with open(video_path, "rb") as video_file:
+        video_bytes = video_file.read()
+        st.video(video_bytes, autoplay=autoplay)
+
+    download_waveform_video(str(video_path))
+
+
 def download_waveform_video(video_path: str):
     """Download video with waveform"""
     gen_video, _ = st.columns(2)

app/uis/audioui.py

@@ -19,4 +19,4 @@ async def audioui(session_id: str, uichat: DeltaGenerator):
         await use_audiocast_request(session_id, summary, content_category)
     else:
         st.info("Audiocast generation completed!")
-        render_audiocast(session_id)
+        await render_audiocast(session_id)