init

Signed-off-by: bitliu <[email protected]>

app.py

@@ -180,56 +180,80 @@ def main():
         st.markdown("- [Models](https://huggingface.co/LLM-Semantic-Router)")
         st.markdown("- [GitHub](https://github.com/vllm-project/semantic-router)")
 
+    # Initialize session state
+    if "text_input" not in st.session_state:
+        st.session_state.text_input = ""
+    if "result" not in st.session_state:
+        st.session_state.result = None
+
     # Main content
-    col1, col2 = st.columns([2, 1])
+    st.subheader("📝 Input")
+    text_input = st.text_area("Enter text to analyze:", height=120, placeholder="Type your text here...", key="input_area")
 
-    with col1:
-        st.subheader("Input")
-        selected_example = st.selectbox("Try an example:", ["Custom input..."] + model_config["examples"])
-        if selected_example == "Custom input...":
-            text_input = st.text_area("Enter text to analyze:", height=120, placeholder="Type your text here...")
+    # Examples section - clickable buttons
+    st.markdown("**💡 Try an example:**")
+    example_cols = st.columns(len(model_config["examples"]))
+    for i, (col, example) in enumerate(zip(example_cols, model_config["examples"])):
+        with col:
+            if st.button(f"Example {i+1}", key=f"ex_{i}", use_container_width=True, help=example):
+                st.session_state.text_input = example
+                st.rerun()
+
+    # Show example text preview
+    st.caption("Hover over buttons to preview examples")
+
+    st.markdown("---")
+
+    # Analyze button
+    if st.button("🔍 Analyze", type="primary", use_container_width=True):
+        if not text_input.strip():
+            st.warning("Please enter some text to analyze.")
         else:
-            text_input = st.text_area("Enter text to analyze:", value=selected_example, height=120)
+            with st.spinner("Analyzing..."):
+                if model_config["type"] == "sequence":
+                    label, emoji, conf, scores = classify_sequence(
+                        text_input, model_config["id"], model_config["labels"]
+                    )
+                    st.session_state.result = {
+                        "type": "sequence",
+                        "label": label,
+                        "emoji": emoji,
+                        "confidence": conf,
+                        "scores": scores
+                    }
+                else:
+                    entities = classify_tokens(text_input, model_config["id"])
+                    st.session_state.result = {
+                        "type": "token",
+                        "entities": entities,
+                        "text": text_input
+                    }
 
-    with col2:
-        st.subheader("Results")
-        if st.button("🔍 Analyze", type="primary", use_container_width=True):
-            if not text_input.strip():
-                st.warning("Please enter some text to analyze.")
+    # Display results
+    if st.session_state.result:
+        st.markdown("---")
+        st.subheader("📊 Results")
+        result = st.session_state.result
+        if result["type"] == "sequence":
+            col1, col2 = st.columns([1, 1])
+            with col1:
+                st.success(f"{result['emoji']} **{result['label']}**")
+                st.metric("Confidence", f"{result['confidence']:.1%}")
+            with col2:
+                st.markdown("**All Scores:**")
+                sorted_scores = dict(sorted(result["scores"].items(), key=lambda x: x[1], reverse=True))
+                for k, v in sorted_scores.items():
+                    st.progress(v, text=f"{k}: {v:.1%}")
+        else:
+            entities = result["entities"]
+            if entities:
+                st.success(f"Found {len(entities)} PII entity(s)")
+                for e in entities:
+                    st.markdown(f"- **{e['type']}**: `{e['text']}`")
+                st.markdown("### Highlighted Text")
+                components.html(create_highlighted_html(result["text"], entities), height=150)
             else:
-                with st.spinner("Analyzing..."):
-                    if model_config["type"] == "sequence":
-                        label, emoji, conf, scores = classify_sequence(
-                            text_input, model_config["id"], model_config["labels"]
-                        )
-                        st.success(f"{emoji} **{label}**")
-                        st.metric("Confidence", f"{conf:.1%}")
-                        with st.expander("All scores"):
-                            sorted_scores = dict(sorted(scores.items(), key=lambda x: x[1], reverse=True))
-                            for k, v in sorted_scores.items():
-                                st.progress(v, text=f"{k}: {v:.1%}")
-                    else:
-                        entities = classify_tokens(text_input, model_config["id"])
-                        if entities:
-                            st.success(f"Found {len(entities)} PII entity(s)")
-                            for e in entities:
-                                st.markdown(f"- **{e['type']}**: `{e['text']}`")
-                        else:
-                            st.info("✅ No PII detected")
-
-    # Show highlighted text for NER
-    if model_config["type"] == "token" and text_input.strip():
-        if "last_ner_input" in st.session_state and st.session_state.last_ner_input == text_input:
-            st.markdown("### Highlighted Text")
-            components.html(create_highlighted_html(text_input, st.session_state.last_entities), height=150)
-
-    # Store NER results for display
-    if st.button("🔍 Analyze", key="hidden", disabled=True, type="secondary"):
-        pass  # Placeholder
-    if model_config["type"] == "token" and text_input.strip():
-        entities = classify_tokens(text_input, model_config["id"])
-        st.session_state.last_ner_input = text_input
-        st.session_state.last_entities = entities
+                st.info("✅ No PII detected")
 
     # Footer
     st.markdown("---")