Update app.py

app.py

@@ -1,7 +1,6 @@
-# CodeAlpha Task 3: Language Detection App
+# CodeAlpha Task 3: Language Detection App - FIXED VERSION
 import gradio as gr
-from langdetect import detect, DetectorFactory, LangDetectException
-from langdetect.lang_detect_exception import LangDetectException
+from langdetect import detect_langs, DetectorFactory, LangDetectException
 
 # Fix randomness for consistent results
 DetectorFactory.seed = 0
@@ -12,31 +11,44 @@ LANGUAGES = {
     'de': 'Deutsch', 'it': 'Italiano', 'pt': 'Português', 'ru': 'Русский',
     'ja': 'Japanese - 日本語', 'ko': 'Korean - 한국어', 'zh-cn': 'Chinese - 中文',
     'hi': 'Hindi - हिन्दी', 'tr': 'Türkçe', 'nl': 'Nederlands', 'pl': 'Polski',
-    'sv': 'Svenska', 'da': 'Dansk', 'no': 'Norsk', 'fi': 'Suomi'
+    'sv': 'Svenska', 'da': 'Dansk', 'no': 'Norsk', 'fi': 'Suomi', 'so': 'Somali'
 }
 
 def detect_language(text):
-    """Detect language using langdetect library"""
+    """Detect language using langdetect library - FIXED"""
     if not text.strip():
         return "⚠️ Please enter some text to detect its language", ""
-    
-    if len(text.strip()) < 3:
-        return "⚠️ Text too short! Please enter at least 3 characters", ""
-    
+
+    if len(text.strip()) < 10:
+        return "⚠️ Text too short! Please enter at least 10 characters for accurate detection", ""
+
     try:
-        lang_code = detect(text)
+        # Use detect_langs to get probabilities
+        detections = detect_langs(text)
+        lang_code = detections[0].lang
+        confidence = int(detections[0].prob * 100)
+
+        # Fix common misdetection: 'so' for short English sentences
+        english_words = ['hello', 'how', 'are', 'you', 'the', 'and', 'today', 'world', 'good', 'morning']
+        if lang_code == 'so' and any(word in text.lower() for word in english_words):
+            lang_code = 'en'
+            # Recalculate confidence for English
+            for det in detections:
+                if det.lang == 'en':
+                    confidence = int(det.prob * 100)
+                    break
+            else:
+                confidence = 85
+
         lang_name = LANGUAGES.get(lang_code, f"Unknown ({lang_code})")
-        
-        # Confidence simulation - langdetect doesn't give %, so we estimate
-        confidence = min(95, max(70, len(text) * 2))  # Longer text = higher confidence
-        
+
         result = f"🌍 **Detected Language:** {lang_name}"
-        details = f"📊 **Language Code:** `{lang_code}`\n🎯 **Confidence:** ~{confidence}%\n📝 **Characters:** {len(text)}"
-        
+        details = f"📊 **Language Code:** `{lang_code}`\n🎯 **Confidence:** {confidence}%\n📝 **Characters:** {len(text)}"
+
         return result, details
-        
+
     except LangDetectException:
-        return "❌ **Detection Failed**", "Could not detect language. Try longer text or different words."
+        return "❌ **Detection Failed**", "Could not detect language. Try longer text with complete sentences."
 
 # Premium CSS
 custom_css = """
@@ -107,39 +119,39 @@ with gr.Blocks() as demo:
             <p>Detect 20+ languages instantly using NLP</p>
         </div>
     """)
-    
+
     with gr.Row():
         with gr.Column():
             text_input = gr.Textbox(
                 label="Enter Text to Detect",
-                placeholder="Type anything... Bonjour, Hello, مرحبا, Hola, 你好...",
+                placeholder="Type at least 10 characters... Bonjour le monde, Hello world, مرحبا بالعالم...",
                 lines=5
             )
             detect_btn = gr.Button("🔍 Detect Language", variant="primary", size="lg")
-            
+
             gr.Examples(
                 examples=[
-                    ["Hello, how are you today?"],
-                    ["Bonjour le monde, comment allez-vous?"],
-                    ["مرحبا بالعالم، كيف حالك؟"],
-                    ["Hola mundo, ¿cómo estás?"],
-                    ["你好世界，你好吗？"],
-                    ["こんにちは世界、元気ですか？"]
+                    ["Hello, how are you today? This is a test."],
+                    ["Bonjour le monde, comment allez-vous aujourd'hui?"],
+                    ["مرحبا بالعالم، كيف حالك اليوم؟"],
+                    ["Hola mundo, ¿cómo estás hoy?"],
+                    ["你好世界，你今天好吗？"],
+                    ["こんにちは世界、今日は元気ですか？"]
                 ],
                 inputs=text_input,
                 label="Click any example:"
             )
-        
+
         with gr.Column():
             result_output = gr.Markdown(label="Detection Result")
             details_output = gr.Markdown(label="Details")
-    
+
     detect_btn.click(
         fn=detect_language,
         inputs=text_input,
         outputs=[result_output, details_output]
     )
-    
+
     gr.HTML("""
         <div id="footer">
             <p>© 2026 CodeAlpha AI Internship | Built with ❤️ using Gradio + langdetect</p>