Update app.py

app.py

@@ -176,27 +176,27 @@ def predict(text_input, progress=gr.Progress()):
     print(records_df)
     
     
-    progress(0.2, desc="Embedding Data...")
+    progress(0.3, desc="Embedding Data...")
     texts_to_embedd = [title + tokenizer.sep_token + publication + tokenizer.sep_token  + abstract for title, publication, abstract in zip(records_df['title'],records_df['parsed_publication'], records_df['abstract'])]
     
     embeddings = create_embeddings(texts_to_embedd)
     print(embeddings)
 
-    progress(0.3, desc="Project into UMAP-embedding...")
+    progress(0.5, desc="Project into UMAP-embedding...")
     umap_embeddings = mapper.transform(embeddings)
     records_df[['x','y']] = umap_embeddings
 
     basedata_df['color'] = '#ced4d211'
     records_df['color'] = '#a81a26'
 
-    progress(0.4, desc="Set up data...")
+    progress(0.6, desc="Set up data...")
 
-    stacked_df = pd.concat([records_df, basedata_df], axis=0, ignore_index=True)
+    stacked_df = pd.concat([basedata_df,records_df], axis=0, ignore_index=True)
     stacked_df = stacked_df.fillna("Unlabelled")
     stacked_df = stacked_df.reset_index(drop=True)
     print(stacked_df)
 
-    
+    extra_data = pd.DataFrame(stacked_df['doi'])
     
     
     file_name = f"{datetime.utcnow().strftime('%s')}.html"
@@ -205,16 +205,76 @@ def predict(text_input, progress=gr.Progress()):
  
    #
     
-    progress(0.5, desc="Plotting...")
+    progress(0.7, desc="Plotting...")
+
+    custom_css = """
+    
+       
+            #title-container {
+                background: #edededaa;
+                border-radius: 2px;
+    
+                box-shadow: 2px 3px 10px #aaaaaa00;
+            }
+            
+    
+            
+        #search-container {
+            position: fixed !important;
+            top: 20px !important;
+            right: 20px !important;
+            left: auto !important;
+            width: 200px !important;
+            z-index: 9999 !important;
+        }
+        
+        #search {
+         //   padding: 8px 8px !important;
+         //   border: none !important;
+         //   border-radius: 20px !important;
+            background-color: #ffffffaa !important;
+            font-family: 'Roboto Condensed', sans-serif !important;
+            font-size: 14px;
+         //   box-shadow: 0 0px 0px #aaaaaa00 !important;
+        }
+    
+            
+    
+    """
 
+    
     plot = datamapplot.create_interactive_plot(
         stacked_df[['x','y']].values,
         np.array(stacked_df['cluster_1_labels']),np.array(stacked_df['cluster_2_labels']),np.array(stacked_df['cluster_3_labels']),
         hover_text=[str(ix) + ', ' + str(row['parsed_publication']) + str(row['title']) for ix, row in stacked_df.iterrows()],
-        font_family="Roboto Condensed",marker_color_array=stacked_df['color']
-    )
+        marker_color_array=stacked_df['color'],
+
+        use_medoids=True,
+        width=1100,
+        height=900,
+    #    title='The Science of <span style="color:#ab0b00;"> Consciousness </span>',
+    #    sub_title=f'<div style="margin-top:20px;"> Large sample, n={len(dataset_df_filtered)}, embeddings with specter 2 & UMAP, labels: Claude 3.5 Sonnet </div>',
+        point_radius_min_pixels=1,
+        text_outline_width=5,
+        point_hover_color='#ab0b00',
+        point_radius_max_pixels=7,
+        color_label_text=False,
+        font_family="Roboto Condensed",
+        font_weight=700,
+        tooltip_font_weight=600,
+        tooltip_font_family="Roboto Condensed",
+        extra_point_data=extra_data,
+        on_click="window.open(`{doi}`)",
+        custom_css=custom_css,
+        initial_zoom_fraction=.5,
+        enable_search=True)
+
+
+
+
+    
 
-    progress(0.6, desc="Saving plot...")
+    progress(0.9, desc="Saving plot...")
     plot.save(file_path)
 
     progress(1.0, desc="Done!")
@@ -222,6 +282,15 @@ def predict(text_input, progress=gr.Progress()):
     link = f'<a href="/static/{file_name}" target="_blank">{file_name}</a>'
     return link, iframe
 
+
+
+
+
+
+################ MAIN BLOCK #####################
+
+
+
 with gr.Blocks() as block:
     gr.Markdown("""
 ## Gradio + FastAPI + Static Server
@@ -240,6 +309,14 @@ The Gradio app generates dynamic HTML files and stores them in a static director
 
 
 
+
+
+
+
+
+
+
+
 def setup_basemap_data():
     # get data.
     print("getting basemap data...")