Rename app_10.py to app_11.py

app_10.py → app_11.py

@@ -174,6 +174,15 @@ def mark_text_by_position(original_text, start_idx, end_idx, color):
         )
     return original_text  # Return unchanged if indices are invalidt  # Return unchanged text if no span is found
 
+def mark_text_by_tokens(tokenizer, tokens, start_idx, end_idx, color):
+    """Highlights a span in tokenized text using HTML."""
+    highlighted_tokens = copy.deepcopy(tokens)  # Avoid modifying original tokens
+    if start_idx is not None and end_idx is not None and start_idx < end_idx:
+        highlighted_tokens[start_idx] = f"<span style='background-color:{color}; padding:2px; border-radius:4px;'>{highlighted_tokens[start_idx]}"
+        highlighted_tokens[end_idx] = f"{highlighted_tokens[end_idx]}</span>"
+    return tokenizer.convert_tokens_to_string(highlighted_tokens)
+
+
 st.title("Causal Relation Extraction")
 input_text = st.text_area("Enter your text here:", height=300)
 beam_search = st.radio("Enable Beam Search?", ('No', 'Yes')) == 'Yes'
@@ -183,28 +192,26 @@ if st.button("Extract1"):
     if input_text:
         start_cause1, end_cause1, start_cause2, end_cause2, start_effect1, end_effect1, start_effect2, end_effect2, start_signal, end_signal = extract_arguments(input_text, tokenizer, model, beam_search=beam_search)
 
-        cause_text1 = mark_text_by_position(input_text, start_cause1, end_cause1, "#FFD700")  # Gold for cause
-        effect_text1 = mark_text_by_position(input_text, start_effect1, end_effect1, "#90EE90")  # Light green for effect
-        signal_text = mark_text_by_position(input_text, start_signal, end_signal, "#FF6347")  # Tomato red for signal
-
-        st.markdown(f"<span style='font-size: 24px;'><strong>Relation 1:</strong></span>", unsafe_allow_html=True)
-        st.markdown(f"**Cause:**<br>{cause_text1}", unsafe_allow_html=True)
-        st.markdown(f"**Effect:**<br>{effect_text1}", unsafe_allow_html=True)
-        st.markdown(f"**Signal:**<br>{signal_text}", unsafe_allow_html=True)
+        # Convert text to tokenized format
+        tokenized_input = tokenizer.tokenize(input_text)
+        
+        cause_text1 = mark_text_by_tokens(tokenizer, tokenized_input, start_cause1, end_cause1, "#FFD700")  # Gold for cause
+        effect_text1 = mark_text_by_tokens(tokenizer, tokenized_input, start_effect1, end_effect1, "#90EE90")  # Light green for effect
+        signal_text = mark_text_by_tokens(tokenizer, tokenized_input, start_signal, end_signal, "#FF6347")  # Tomato red for signal
 
-        #st.write("List 1:", list1)
+        # Display first relation
+        st.markdown(f"<strong>Relation 1:</strong>", unsafe_allow_html=True)
+        st.markdown(f"**Cause:** {cause_text1}", unsafe_allow_html=True)
+        st.markdown(f"**Effect:** {effect_text1}", unsafe_allow_html=True)
+        st.markdown(f"**Signal:** {signal_text}", unsafe_allow_html=True)
 
+        # Display second relation if beam search is enabled
         if beam_search:
+            cause_text2 = mark_text_by_tokens(tokenizer, tokenized_input, start_cause2, end_cause2, "#FFD700")
+            effect_text2 = mark_text_by_tokens(tokenizer, tokenized_input, start_effect2, end_effect2, "#90EE90")
 
-            cause_text2 = mark_text_by_position(input_text, start_cause2, end_cause2, "#FFD700")  # Gold for cause
-            effect_text2 = mark_text_by_position(input_text, start_effect2, end_effect2, "#90EE90")  # Light green for effect
-            signal_text = mark_text_by_position(input_text, start_signal, end_signal, "#FF6347")  # Tomato red for signal
-    
-            st.markdown(f"<span style='font-size: 24px;'><strong>Relation 2:</strong></span>", unsafe_allow_html=True)
-            st.markdown(f"**Cause:**<br>{cause_text2}", unsafe_allow_html=True)
-            st.markdown(f"**Effect:**<br>{effect_text2}", unsafe_allow_html=True)
-            st.markdown(f"**Signal:**<br>{signal_text}", unsafe_allow_html=True)
-
-            #st.write("List 2:", list2)
+            st.markdown(f"<strong>Relation 2:</strong>", unsafe_allow_html=True)
+            st.markdown(f"**Cause:** {cause_text2}", unsafe_allow_html=True)
+            st.markdown(f"**Effect:** {effect_text2}", unsafe_allow_html=True)
     else:
         st.warning("Please enter some text before extracting.")