Update app.py

app.py

@@ -46,8 +46,8 @@ table = db.open_table("edunat19")
 def hybrid_search(text):
     results = table.search(text, query_type="hybrid").limit(5).to_pandas()
     
-    # Add a check for duplicate hashes
-    seen_hashes = set()
+    # Use a list to maintain order
+    seen_hashes = []
     
     document = []
     document_html = []
@@ -58,17 +58,26 @@ def hybrid_search(text):
         if hash_id in seen_hashes:
             continue
             
-        seen_hashes.add(hash_id)
+        seen_hashes.append(hash_id)  # append instead of add to maintain order
         title = row['section']
         content = row['text']
 
-        document.append(f"<|source_start|><|source_id_start|>{hash_id}<|source_id_end|>{title}\n{content}<|source_end|>")
+        source_text = f"<|source_start|><|source_id_start|>{hash_id}<|source_id_end|>{title}\n{content}<|source_end|>"
+        document.append(source_text)
         document_html.append(f'<div class="source" id="{hash_id}"><p><b>{hash_id}</b> : {title}<br>{content}</div>')
 
+        # Print for debugging
+        print(f"Added source {hash_id}")
+        print(f"Length of source text: {len(source_text)}")
+
     document = "\n".join(document)
     document_html = '<div id="source_listing">' + "".join(document_html) + "</div>"
+    
+    # Print total length for debugging
+    print(f"Total length of document: {len(document)}")
+    
     return document, document_html
-
+    
 class pleiasBot:
     def __init__(self, system_prompt="Tu es Appli, un asistant de recherche qui donne des responses sourcées"):
         self.system_prompt = system_prompt