Update app.py

app.py

@@ -8,6 +8,21 @@ import os
 import io
 
 
+# Initialize session state for storing data
+if 'scenario' not in st.session_state:
+    st.session_state.scenario = None
+if 'scenario_zh' not in st.session_state:
+    st.session_state.scenario_zh = None
+if 'story' not in st.session_state:
+    st.session_state.story = None
+if 'story_zh' not in st.session_state:
+    st.session_state.story_zh = None
+if 'audio_generated' not in st.session_state:
+    st.session_state.audio_generated = False
+if 'audio_data' not in st.session_state:
+    st.session_state.audio_data = None
+
+
 # function part
 # img2text
 def img2text(url):
@@ -22,6 +37,7 @@ def translate_to_chinese(text):
     translation = translator(text)[0]["translation_text"]
     return translation
 
+
 # text2story
 def text2story(text):
     # Initialize the text generation pipeline
@@ -261,59 +277,92 @@ with st.container():
     uploaded_file = st.file_uploader("", key="upload")
 
 if uploaded_file is not None:
+    # Save uploaded file
     bytes_data = uploaded_file.getvalue()
-    with open(uploaded_file.name, "wb") as file:
+    temp_file_path = uploaded_file.name
+    with open(temp_file_path, "wb") as file:
         file.write(bytes_data)
 
     # Display image
     st.image(uploaded_file, use_column_width=True)
 
-    # Stage 1: Image to Text
-    with st.container():
-        st.markdown("<h3><span class='stage-icon'>🔍</span> 圖片解讀中</h3>", unsafe_allow_html=True)
-        scenario = img2text(uploaded_file.name)
-
-        # Display English caption
-        st.text("英文描述: " + scenario)
-
-        # Translate the caption to Chinese
-        scenario_zh = translate_to_chinese(scenario)
-
-        # Display Chinese caption
-        st.text("中文描述: " + scenario_zh)
-
-    # Stage 2: Text to Story
-    with st.container():
-        st.markdown("<h3><span class='stage-icon'>📝</span> 故事創作中</h3>", unsafe_allow_html=True)
-        story = text2story(scenario)
-        # Display English story
-        st.text("英文故事: " + story)
-        # Translate the story to Chinese
-        story_zh = translate_to_chinese(story)
-        # Display Chinese story
-        st.text("中文故事: " + story_zh)
+    # Reset session state if a new file is uploaded (detect by checking if there's no scenario yet)
+    if st.session_state.scenario is None:
+        # Stage 1: Image to Text
+        with st.container():
+            st.markdown("<h3><span class='stage-icon'>🔍</span> 圖片解讀中</h3>", unsafe_allow_html=True)
+            
+            # Generate caption if not already done
+            st.session_state.scenario = img2text(temp_file_path)
+            
+            # Display English caption
+            st.text("英文描述: " + st.session_state.scenario)
+            
+            # Translate the caption to Chinese
+            st.session_state.scenario_zh = translate_to_chinese(st.session_state.scenario)
+            
+            # Display Chinese caption
+            st.text("中文描述: " + st.session_state.scenario_zh)
+
+        # Stage 2: Text to Story
+        with st.container():
+            st.markdown("<h3><span class='stage-icon'>📝</span> 故事創作中</h3>", unsafe_allow_html=True)
+            
+            # Generate story if not already done
+            st.session_state.story = text2story(st.session_state.scenario)
+            
+            # Display English story
+            st.text("英文故事: " + st.session_state.story)
+            
+            # Translate the story to Chinese
+            st.session_state.story_zh = translate_to_chinese(st.session_state.story)
+            
+            # Display Chinese story
+            st.text("中文故事: " + st.session_state.story_zh)
+    else:
+        # Display saved results from session state
+        with st.container():
+            st.markdown("<h3><span class='stage-icon'>🔍</span> 圖片解讀中</h3>", unsafe_allow_html=True)
+            st.text("英文描述: " + st.session_state.scenario)
+            st.text("中文描述: " + st.session_state.scenario_zh)
+
+        with st.container():
+            st.markdown("<h3><span class='stage-icon'>📝</span> 故事創作中</h3>", unsafe_allow_html=True)
+            st.text("英文故事: " + st.session_state.story)
+            st.text("中文故事: " + st.session_state.story_zh)
 
     # Stage 3: Story to Audio data
     with st.container():
         st.markdown("<h3><span class='stage-icon'>🔊</span> 故事準備朗讀中</h3>", unsafe_allow_html=True)
 
-        # Create audio for Chinese story
-        audio_data = text2audio(story_zh, lang='zh')
-
         # Play button with Cantonese text
         if st.button("🔊 播放故事"):
-            if audio_data:
-                st.audio(audio_data['audio'],
-                         format="audio/wav",
-                         start_time=0,
-                         sample_rate=audio_data['sampling_rate'])
+            # Only generate audio if not already done
+            if not st.session_state.audio_generated:
+                st.session_state.audio_data = text2audio(st.session_state.story_zh, lang='zh')
+                st.session_state.audio_generated = True
+                
+            # Play the audio
+            if st.session_state.audio_data:
+                st.audio(st.session_state.audio_data['audio'],
+                        format="audio/wav",
+                        start_time=0,
+                        sample_rate=st.session_state.audio_data['sampling_rate'])
             else:
                 st.error("哎呀！再試多次啦！")
 
-    # Cleanup: Remove the temporary file
-    if os.path.exists(uploaded_file.name):
-        os.remove(uploaded_file.name)
+    # Cleanup: Remove the temporary file when the user is done
+    if os.path.exists(temp_file_path):
+        os.remove(temp_file_path)
 else:
+    # Clear session state when no file is uploaded
+    st.session_state.scenario = None
+    st.session_state.scenario_zh = None
+    st.session_state.story = None
+    st.session_state.story_zh = None
+    st.session_state.audio_generated = False
+    st.session_state.audio_data = None
+    
     # Welcome message in Cantonese
     st.markdown("""
     <div class="welcome-message">