Update app.py

app.py

@@ -2,14 +2,25 @@ import gradio as gr
 import pandas as pd
 import difflib
 
-# Load your CSV data
-df = pd.read_csv("dialect_data.csv")
+# Load your CSV data with error handling
+try:
+    df = pd.read_csv("dialect_data.csv")
+    print("CSV loaded successfully!")
+    print("Available columns:", df.columns.tolist())
+except Exception as e:
+    print(f"Error loading CSV: {e}")
+    # Create empty dataframe as fallback
+    df = pd.DataFrame(columns=['Dialect Bengali', 'Translation in Ac English', 'Meaning'])
 
 # Function to translate using your CSV data
 def translate_text(text):
     if not text.strip():
         return "Please enter a phrase or question"
     
+    # Check if dataframe is empty
+    if df.empty:
+        return "Database not loaded. Please check your CSV file."
+    
     # Normalize text
     def normalize_phrase(phrase):
         return ''.join(char for char in phrase.lower() if char.isalpha() or char.isspace())
@@ -19,9 +30,17 @@ def translate_text(text):
     # Get all phrases from CSV for matching
     all_phrases = []
     for index, row in df.iterrows():
-        csv_phrase = str(row['Dialect Bengali'])
-        clean_csv = normalize_phrase(csv_phrase)
-        all_phrases.append((clean_csv, row['Translation in Ac English'], row['Meaning'], csv_phrase))
+        try:
+            csv_phrase = str(row['Dialect Bengali'])
+            clean_csv = normalize_phrase(csv_phrase)
+            translation = str(row['Translation in Ac English'])
+            meaning = str(row['Meaning'])
+            all_phrases.append((clean_csv, translation, meaning, csv_phrase))
+        except:
+            continue
+    
+    if not all_phrases:
+        return "No phrases found in database. Check CSV format."
     
     # 1. First try exact match
     for clean_csv, translation, meaning, original in all_phrases:
@@ -33,7 +52,7 @@ def translate_text(text):
         user_clean, 
         [clean_csv for clean_csv, trans, meaning, orig in all_phrases], 
         n=3,
-        cutoff=0.6  # Higher threshold for good matches
+        cutoff=0.6
     )
     
     if close_matches:
@@ -42,55 +61,28 @@ def translate_text(text):
             for clean_csv, translation, meaning, original in all_phrases:
                 if clean_csv == match:
                     similarity = int(difflib.SequenceMatcher(None, user_clean, clean_csv).ratio() * 100)
-                    if similarity >= 60:  # Good matches
+                    if similarity >= 60:
                         suggestions.append(f"🎯 '{original}' ({similarity}% match)\n💬 {translation}\n📖 {meaning}")
         
         if suggestions:
             return "🔍 SIMILAR PHRASES FOUND:\n\n" + "\n\n".join(suggestions)
     
-    # 3. Try substring matching for partial matches
-    partial_matches = []
-    for clean_csv, translation, meaning, original in all_phrases:
-        if user_clean in clean_csv or clean_csv in user_clean:
-            similarity = int(difflib.SequenceMatcher(None, user_clean, clean_csv).ratio() * 100)
-            partial_matches.append((original, translation, meaning, similarity))
-    
-    if partial_matches:
-        # Sort by similarity score
-        partial_matches.sort(key=lambda x: x[3], reverse=True)
-        best_match = partial_matches[0]
-        return f"🔍 PARTIAL MATCH:\n💬 '{best_match[0]}' -> {best_match[1]}\n📖 {best_match[2]}\n⚡ {best_match[3]}% similar"
-    
-    # 4. Very loose matching for completely different questions
-    loose_matches = difflib.get_close_matches(
-        user_clean, 
-        [clean_csv for clean_csv, trans, meaning, orig in all_phrases], 
-        n=2,
-        cutoff=0.3  # Very low threshold
-    )
-    
-    if loose_matches:
-        return "❓ COMPLETELY DIFFERENT QUESTION\n\n📝 Try asking about:\n" + \
-               "\n".join([f"• '{orig}'" for clean_csv, trans, meaning, orig in all_phrases 
-                         if clean_csv in loose_matches][:3]) + \
-               "\n\n💡 Or ask something related to daily conversations"
-    
-    # 5. Final fallback - show sample phrases
+    # 3. Show sample available phrases
     sample_phrases = [orig for clean_csv, trans, meaning, orig in all_phrases[:5]]
     return "❓ ASK DIFFERENT QUESTION\n\n" + \
-           "📋 Available phrases:\n" + \
+           "📋 Try these phrases:\n" + \
            "\n".join([f"• '{phrase}'" for phrase in sample_phrases]) + \
-           "\n\n💡 Try questions about greetings, daily activities, or time"
+           "\n\n💡 Example: 'Kita kobor?', 'Goto kali', 'gesle ni'"
 
 # Create the app interface
 demo = gr.Interface(
     fn=translate_text,
     inputs=gr.Textbox(label="Type phrase in our language", 
-                     placeholder="Example: gesle ni, Kita kobor?, Goto kali..."),
+                     placeholder="Example: Kita kobor?, Goto kali, gesle ni..."),
     outputs=gr.Textbox(label="Translation Result"),
     title="🌍 Smart Dialect Translator",
-    description="Translates with smart matching - finds similar phrases and suggestions",
-    examples=[["gesle ni"], ["Kita kobor?"], ["Goto kali"], ["hello"], ["how are you"]]
+    description="Translates with smart matching - finds similar phrases",
+    examples=[["gesle ni"], ["Kita kobor?"], ["Goto kali"]]
 )
 
 # Launch the app