Update app.py

app.py

@@ -22,17 +22,33 @@ st.markdown("""
     </style>
     """, unsafe_allow_html=True)
 
-# Define allowed characters
-ALLOWED_CHARS = set('4O892ERSZPBFVQWXYACIGH1TU0DNM3JKL567(n)(v)')
+# Define allowed characters (single characters and multi-character tokens)
+ALLOWED_SINGLE_CHARS = set('4O892ERSZPBFVQWXYACIGH1TU0DNM3JKL567')
+ALLOWED_MULTI_CHARS = ['(n)', '(v)']
 
 def parse_voynich_word(word):
-    """Parse a Voynich word into individual characters - filtering to allowed characters only"""
+    """Parse a Voynich word into individual characters - treating (n) and (v) as single units"""
     if not word or word.strip() == '':
         return None, None
     
     word = word.strip()
-    # Filter to only allowed characters
-    chars = [c for c in list(word) if c in ALLOWED_CHARS]
+    chars = []
+    i = 0
+    
+    while i < len(word):
+        # Check for multi-character tokens first
+        if i + 2 < len(word):
+            three_char = word[i:i+3]
+            if three_char in ALLOWED_MULTI_CHARS:
+                chars.append(three_char)
+                i += 3
+                continue
+        
+        # Otherwise check single character
+        if word[i] in ALLOWED_SINGLE_CHARS:
+            chars.append(word[i])
+        
+        i += 1
     
     # If no valid characters remain, return None
     if not chars:
@@ -179,7 +195,6 @@ def get_download_link_csv(df, filename):
 
 st.title("Voynich Manuscript Analyzer")
 st.write("Upload your CSV file.")
-st.info(f"**Filtered Character Set:** {' '.join(sorted(ALLOWED_CHARS))}")
 
 # Upload eva legend to sidebar
 floating_image_file = st.sidebar.file_uploader("Upload an image",