fix: eliminate redundant initialization of AssetSelector, APIClients, and logging

- Remove duplicate logging.basicConfig() in setup_gcs_permissions.py
- Add SHARED_ASSET_SELECTOR and SHARED_API_CLIENTS globals in process_csv.py
- Modify ContentAutomation to accept optional asset_selector and api_clients params
- Update data_holder reference before each row to maintain state consistency
- Remove duplicate ON_SCREEN_TEXT check in process_row()

This reduces video/audio library loads from 3x to 1x per job and GCS/TTS client
initialization from 2x to 1x per job.

src/automation.py

@@ -28,12 +28,14 @@ from file_downloader import FileDownloader
 from data_holder import DataHolder
 
 class ContentAutomation:
-    def __init__(self, config: Dict[str, Any], data_holder: DataHolder = None):
+    def __init__(self, config: Dict[str, Any], data_holder: DataHolder = None, asset_selector: 'AssetSelector' = None, api_clients: 'APIClients' = None):
         self.config = config
         self.data_holder = data_holder or DataHolder()
-        self.api_clients = APIClients(config, self.data_holder)
+        # Reuse provided api_clients or create new one
+        self.api_clients = api_clients or APIClients(config, self.data_holder)
         self.video_renderer = VideoRenderer(config, self.data_holder)
-        self.asset_selector = AssetSelector(config, self.data_holder)
+        # Reuse provided asset_selector or create new one
+        self.asset_selector = asset_selector or AssetSelector(config, self.data_holder)
         self.file_downloader = FileDownloader()
         self.pipeline_start_time = None
 

src/google_src/setup_gcs_permissions.py

@@ -9,9 +9,11 @@ from google.cloud import storage
 from google.iam.v1 import policy_pb2
 from google_src.gcloud_wrapper import get_default_wrapper
 
-# Configure logging
-logging.basicConfig(level=logging.INFO, format='%(asctime)s - %(levelname)s - %(message)s')
-logger = logging.getLogger(__name__)
+# Use the project's configured logger to avoid duplicate log output
+try:
+    from utils import logger
+except ImportError:
+    logger = logging.getLogger(__name__)
 
 def setup_bucket_permissions(bucket_name: str, members: list, role: str = "roles/storage.objectViewer", storage_client=None):
     """

src/process_csv.py

@@ -9,6 +9,7 @@ from main import (
     run_pipeline,
 )
 from automation import ContentAutomation
+from api_clients import APIClients
 from utils import logger
 from data_holder import DataHolder
 from asset_selector import AssetSelector
@@ -20,6 +21,8 @@ from google_src.gcs_utils import list_gcs_files
 
 DATA_DIR = Path("data")
 ALL_VIDEO_FILE_INFO = None
+SHARED_ASSET_SELECTOR = None  # Shared instance to avoid redundant sheet loads
+SHARED_API_CLIENTS = None  # Shared instance to avoid redundant GCS/TTS client initialization
 
 
 def load_executed_from_gsheet(setup_type=None, job_index=None):
@@ -97,19 +100,28 @@ def log_progress_to_gsheet(tts_script: str, result: dict, job_index: int, commit
 
 async def process_row(row, config: dict):
     """Process one CSV row using the main pipeline."""
-    global ALL_VIDEO_FILE_INFO
+    global ALL_VIDEO_FILE_INFO, SHARED_ASSET_SELECTOR, SHARED_API_CLIENTS
     tts_script = row.get("TTS Script (AI Avatar)", "")
     if os.getenv("ON_SCREEN_TEXT", "false").lower() == "true":
         tts_script = row.get("On-Screen Text", "").strip()
-    if os.getenv("ON_SCREEN_TEXT", "false").lower() == "true":
-        tts_script = row.get("On-Screen Text", "").strip()
 
     logger.info(f"▶️ Executing: {tts_script}...")
 
     dataHolder = DataHolder()
     dataHolder.visual_assets["all_videos"] = ALL_VIDEO_FILE_INFO
 
-    automation = ContentAutomation(config, dataHolder)
+    # Update shared instances with current dataHolder before use
+    if SHARED_ASSET_SELECTOR:
+        SHARED_ASSET_SELECTOR.data_holder = dataHolder
+    if SHARED_API_CLIENTS:
+        SHARED_API_CLIENTS.data_holder = dataHolder
+
+    # Reuse shared AssetSelector and APIClients to avoid redundant initialization
+    automation = ContentAutomation(
+        config, dataHolder,
+        asset_selector=SHARED_ASSET_SELECTOR,
+        api_clients=SHARED_API_CLIENTS
+    )
 
     content_strategy = {
         "gemini_prompt": row.get("Gemini Imagen4 Ultra Prompt (specific)", ""),
@@ -130,23 +142,35 @@ async def process_row(row, config: dict):
 
 
 async def download_all_video(config: dict):
-    """Download all library videos once and cache them."""
-    global ALL_VIDEO_FILE_INFO
+    """Download all library videos once and cache them. Creates shared instances."""
+    global ALL_VIDEO_FILE_INFO, SHARED_ASSET_SELECTOR, SHARED_API_CLIENTS
 
     if ALL_VIDEO_FILE_INFO is None:
         logger.info("📥 Pre-downloading all library videos...")
 
-        asset_selector = AssetSelector(config)
+        # Create the shared AssetSelector once - this loads video/audio libraries from sheets
+        if SHARED_ASSET_SELECTOR is None:
+            SHARED_ASSET_SELECTOR = AssetSelector(config)
+        
         video_urls = [
             row.get("Video URL (No Audio)", "").strip()
-            for _, row in asset_selector.video_library.iterrows()
+            for _, row in SHARED_ASSET_SELECTOR.video_library.iterrows()
             if row.get("Video URL (No Audio)", "").strip()
         ]
 
         dataHolder = DataHolder()
         dataHolder.visual_assets["all_videos"] = [{"url": url} for url in video_urls]
 
-        automation = ContentAutomation(config, dataHolder)
+        # Create the shared APIClients once - this initializes GCS/TTS clients
+        if SHARED_API_CLIENTS is None:
+            SHARED_API_CLIENTS = APIClients(config, dataHolder)
+
+        # Pass the shared instances to avoid creating new ones
+        automation = ContentAutomation(
+            config, dataHolder,
+            asset_selector=SHARED_ASSET_SELECTOR,
+            api_clients=SHARED_API_CLIENTS
+        )
         await automation._download_all_visual_assets()
 
         ALL_VIDEO_FILE_INFO = dataHolder.visual_assets.get("all_videos", [])