triggering feature span caching on precomputed regions

app.py

@@ -428,8 +428,6 @@ def app(share=False):
             label="Precomputed Regions",
             info="Select a region to automatically zoom and analyze"
         )
-        # Add a hidden HTML component for JavaScript execution
-        js_trigger = gr.HTML(visible=False, elem_id="js-trigger")
         
         
         with gr.Row():

precompute_caches.py

@@ -1,3 +1,4 @@
+import ast
 import os
 import json
 import pickle
@@ -8,9 +9,8 @@ from datetime import datetime
 import yaml
 
 # Import your actual modules exactly as app.py does
-from utils.visualizations import get_instances, load_interp_space, compute_tsne_with_cache, compute_precomputed_regions
-from utils.ui import update_task_display, instance_to_df
-from utils.interp_space_utils import cached_generate_style_embedding, compute_g2v_features, compute_predicted_author
+from utils.visualizations import get_instances, load_interp_space, trigger_precomputed_region, handle_zoom_with_retries
+from utils.ui import update_task_display
 
 def load_config(path="config/config.yaml"):
     with open(path, "r") as f:
@@ -132,6 +132,39 @@ def precompute_all_caches(
                 
                 print(f"  ✓ Instance {instance_id} with model {model_name} completed successfully")
                 
+
+                print("  → Testing region zoom simulation...")
+                if precomputed_regions_state:
+                    regions_dict = ast.literal_eval(precomputed_regions_state)
+                    test_regions = list(regions_dict.keys()) 
+                    
+                    for region_name in test_regions:
+                        try:
+                            print(f"    → Testing region: {region_name}")
+                            
+                            # Step 3a: Simulate region selection (trigger_precomputed_region)
+                            zoom_payload = trigger_precomputed_region(region_name, regions_dict)
+                            
+                            if zoom_payload:  # Only proceed if we got a valid zoom payload
+                                # Step 3b: Simulate axis_ranges.change() (handle_zoom_with_retries)
+                                zoom_results = handle_zoom_with_retries(
+                                    event_json=zoom_payload,
+                                    bg_proj=bg_proj,
+                                    bg_lbls=bg_ids,
+                                    clustered_authors_df=background_authors_embeddings_df,
+                                    task_authors_df=task_authors_embeddings_df
+                                )
+                                
+                                # Extract results like app.py does
+                                (features_rb_update, gram2vec_rb_update, llm_style_feats_analysis, 
+                                feature_list_state, visible_zoomed_authors) = zoom_results
+                                
+                                print(f"      ✓ LLM features cached for region: {region_name}")
+                                
+                        except Exception as e:
+                            print(f"      ✗ Failed to cache features for region {region_name}: {e}")
+                            # Continue with other regions even if one fails
+                            continue
             except Exception as e:
                 error_msg = f"Error processing instance {instance_id} with model {model_name}: {str(e)}"
                 print(f"  ✗ {error_msg}")