fix

.streamlit/secrets.toml

@@ -0,0 +1 @@
+ 

app copy.py

@@ -0,0 +1,135 @@
+import streamlit as st
+from st_pages import Page, show_pages
+from openai import OpenAI
+
+from whisper_stt import whisper_stt
+
+
+
+
+# Set page configuration
+st.set_page_config(layout="wide")
+show_pages([Page("app.py", "Home", "🏠")])
+
+# Initialize session state variables
+if 'paused' not in st.session_state:
+    st.session_state.paused = False
+if 'question_text' not in st.session_state:
+    st.session_state.question_text = ""
+if 'submitted' not in st.session_state:
+    st.session_state.submitted = False   
+if 'response_content' not in st.session_state:
+    st.session_state.response_content = ""
+if 'stopped' not in st.session_state:
+    st.session_state.stopped = False
+if 'function_call_count' not in st.session_state:
+    st.session_state.function_call_count = 0       
+if 'transcribed_text' not in st.session_state:
+    st.session_state.transcribed_text = ""
+if 'last_processed_text' not in st.session_state:
+    st.session_state.last_processed_text = ""
+if 'headers' not in st.session_state:
+    st.session_state.headers = []
+
+def create_anchor_link(text):
+    if text is None:
+        return ""
+    return f"<a href='#{text.strip().lower().replace(' ', '-').replace(',', '').replace('.', '').replace(chr(39), '-')}'>{text}</a>"
+
+def on_stop():
+    st.session_state.stopped = True
+
+col0 = st.columns(1)[0]
+placeholder = col0.empty()  # Define the placeholder
+def handle_enter(key):
+    if key == "ctrl+enter":
+        new_question = st.session_state.question_input
+        print(f"handle_enter called. new_question: '{new_question}'")
+        print(f"session state: {st.session_state}")
+with st.sidebar:
+    api_key = st.text_input("API Key", key="chatbot_api_key", type="password")
+
+    col1, col2 = st.columns(2)
+
+    with col1:
+        # Call whisper_stt without a callback
+        transcribed_text = whisper_stt(
+            openai_api_key=api_key,
+            language='en'
+        )
+        if transcribed_text:
+            st.session_state.question_text = transcribed_text
+
+        # Check if new transcription is available
+        if transcribed_text and transcribed_text != st.session_state.transcribed_text:
+            st.session_state.transcribed_text = transcribed_text
+            st.session_state.question_text = transcribed_text
+            st.session_state.submitted = True
+
+    if st.session_state.question_text:
+        st.markdown(create_anchor_link(st.session_state.question_text), unsafe_allow_html=True)  
+
+    if 'question_input' in st.session_state and st.session_state.question_input:
+        st.markdown(create_anchor_link(st.session_state.question_input), unsafe_allow_html=True)         
+    with col2:
+        st.button(label='Stop', on_click=on_stop)
+   
+    # Create an input for the question and use new_question directly
+    new_question = st.text_area("Question", 
+        value=st.session_state.question_text or "", 
+        height=150, 
+        key="question_input",
+        on_change=handle_enter,
+        args=("ctrl+enter",)
+    )
+    print(f"After text_area, new_question: '{new_question}'")
+    # Check if new_question has changed and is not empty
+    if new_question and new_question != st.session_state.question_text:
+        st.session_state.question_text = new_question
+        st.session_state.submitted = True
+
+    st.markdown("## Navigation")
+    for header in st.session_state.headers:
+        st.markdown(create_anchor_link(header), unsafe_allow_html=True)    
+
+if st.session_state.question_text and not api_key:
+    st.info("Please add your OpenAI API key to continue.")
+    st.stop()
+
+if st.session_state.submitted and not st.session_state.stopped:
+    st.session_state.headers.append(st.session_state.question_text)
+    
+    if st.session_state.function_call_count == 0:
+        header = f'## {st.session_state.question_text}'
+        anchor = create_anchor_link(st.session_state.question_text)
+        st.session_state.response_content += f'{header}\n\n{anchor}\n\n'
+    else:
+        header = f'\n\n---\n{st.session_state.question_text}\n---\n\n'
+        anchor = create_anchor_link(st.session_state.question_text)
+        st.session_state.response_content += f'{header}{anchor}\n\n'
+    
+    st.session_state.function_call_count += 1
+    client = OpenAI(api_key=api_key)
+    st.session_state.messages = [{"role": "user", "content": st.session_state.question_text}]
+
+    response = client.chat.completions.create(
+        model="gpt-4o",
+        messages=st.session_state.messages,
+        stream=True
+    )
+
+    for chunk in response:
+        if st.session_state.stopped:
+            st.session_state.stopped = False
+            st.session_state.submitted = False
+            break
+        else:
+            if chunk and chunk.choices[0].delta.content:
+                st.session_state.response_content += chunk.choices[0].delta.content
+                placeholder.markdown(st.session_state.response_content, unsafe_allow_html=True)
+
+    st.session_state.submitted = False
+    st.session_state.stopped = False
+
+placeholder.markdown(st.session_state.response_content, unsafe_allow_html=True)
+st.session_state.stopped = False

app.py

@@ -1,68 +1,8 @@
 import streamlit as st
 from st_pages import Page, show_pages
 from openai import OpenAI
-from streamlit_mic_recorder import mic_recorder
-import io
-
-
-
-
-def whisper_stt(openai_api_key, start_prompt="Start recording", stop_prompt="Stop recording", just_once=False,
-               use_container_width=False, language=None, callback=None, args=(), kwargs=None, key=None):
-    
-    
-
-    if not '_last_speech_to_text_transcript_id' in st.session_state:
-        st.session_state._last_speech_to_text_transcript_id = 0
-    if not '_last_speech_to_text_transcript' in st.session_state:
-        st.session_state._last_speech_to_text_transcript = None
-    if key and not key + '_output' in st.session_state:
-        st.session_state[key + '_output'] = None
-    audio = mic_recorder(start_prompt=start_prompt, stop_prompt=stop_prompt, just_once=just_once,
-                         use_container_width=use_container_width,format="webm", key=key)
-    new_output = False
-    if audio is None:
-        output = None
-    else:
-        if openai_api_key:
-            if not 'openai_client' in st.session_state:
-                #assert openai_api_key, openai_api_key
-                st.session_state.openai_client = OpenAI(api_key=openai_api_key)
-
-            id = audio['id']
-            new_output = (id > st.session_state._last_speech_to_text_transcript_id)
-            if new_output:
-                output = None
-                st.session_state._last_speech_to_text_transcript_id = id
-                audio_bio = io.BytesIO(audio['bytes'])
-                audio_bio.name = 'audio.webm'
-                success = False
-                err = 0
-                while not success and err < 3:  # Retry up to 3 times in case of OpenAI server error.
-                    try:
-                        transcript = st.session_state.openai_client.audio.transcriptions.create(
-                            model="whisper-1",
-                            file=audio_bio,
-                            language=language
-                        )
-                    except Exception as e:
-                        print(str(e))  # log the exception in the terminal
-                        err += 1
-                    else:
-                        success = True
-                        output = transcript.text
-                        st.session_state._last_speech_to_text_transcript = output
-            elif not just_once:
-                output = st.session_state._last_speech_to_text_transcript
-            else:
-                output = None
-        else:
-            output = None
-    if key:
-        st.session_state[key + '_output'] = output
-    if new_output and callback:
-        callback(*args, **(kwargs or {}))
-    return output
+import urllib.parse
+from whisper_stt import whisper_stt
 
 # Set page configuration
 st.set_page_config(layout="wide")
@@ -91,18 +31,22 @@ if 'headers' not in st.session_state:
 def create_anchor_link(text):
     if text is None:
         return ""
-    return f"<a href='#{text.strip().lower().replace(' ', '-').replace(',', '').replace('.', '').replace(chr(39), '-')}'>{text}</a>"
+    anchor = text.strip().lower().replace(' ', '-').replace(',', '').replace('.', '').replace(chr(39), '-')
+    anchor = urllib.parse.quote(anchor)
+    return f"<a href='#{anchor}'>{text}</a>"
 
 def on_stop():
     st.session_state.stopped = True
 
 col0 = st.columns(1)[0]
 placeholder = col0.empty()  # Define the placeholder
+
 def handle_enter(key):
     if key == "ctrl+enter":
         new_question = st.session_state.question_input
         print(f"handle_enter called. new_question: '{new_question}'")
         print(f"session state: {st.session_state}")
+
 with st.sidebar:
     api_key = st.text_input("API Key", key="chatbot_api_key", type="password")
 
@@ -125,6 +69,8 @@ with st.sidebar:
 
     if st.session_state.question_text:
         st.markdown(create_anchor_link(st.session_state.question_text), unsafe_allow_html=True)  
+        # Debug output
+        st.write("Debug: Generated anchor link:", create_anchor_link(st.session_state.question_text))
 
     if 'question_input' in st.session_state and st.session_state.question_input:
         st.markdown(create_anchor_link(st.session_state.question_input), unsafe_allow_html=True)         
@@ -189,4 +135,27 @@ if st.session_state.submitted and not st.session_state.stopped:
     st.session_state.stopped = False
 
 placeholder.markdown(st.session_state.response_content, unsafe_allow_html=True)
-st.session_state.stopped = False
+st.session_state.stopped = False
+
+# Add anchor points for each header
+for header in st.session_state.headers:
+    anchor = header.strip().lower().replace(' ', '-').replace(',', '').replace('.', '').replace(chr(39), '-')
+    anchor = urllib.parse.quote(anchor)
+    st.markdown(f"<div id='{anchor}'></div>", unsafe_allow_html=True)
+
+# Add JavaScript for smooth scrolling and anchor navigation
+st.markdown(
+    """
+    <script>
+    function navigateToAnchor(anchor) {
+        const element = document.getElementById(anchor);
+        if (element) {
+            element.scrollIntoView({behavior: 'smooth'});
+        } else {
+            console.error('Anchor not found:', anchor);
+        }
+    }
+    </script>
+    """,
+    unsafe_allow_html=True
+)

whisper_stt.py

@@ -7,9 +7,9 @@ import os
 
 def whisper_stt(openai_api_key, start_prompt="Start recording", stop_prompt="Stop recording", just_once=False,
                use_container_width=False, language=None, callback=None, args=(), kwargs=None, key=None):
-    if not 'openai_client' in st.session_state:
-        assert openai_api_key, openai_api_key
-        st.session_state.openai_client = OpenAI(api_key=openai_api_key)
+    
+    
+
     if not '_last_speech_to_text_transcript_id' in st.session_state:
         st.session_state._last_speech_to_text_transcript_id = 0
     if not '_last_speech_to_text_transcript' in st.session_state:
@@ -22,34 +22,40 @@ def whisper_stt(openai_api_key, start_prompt="Start recording", stop_prompt="Sto
     if audio is None:
         output = None
     else:
-        id = audio['id']
-        new_output = (id > st.session_state._last_speech_to_text_transcript_id)
-        if new_output:
-            output = None
-            st.session_state._last_speech_to_text_transcript_id = id
-            audio_bio = io.BytesIO(audio['bytes'])
-            audio_bio.name = 'audio.webm'
-            success = False
-            err = 0
-            while not success and err < 3:  # Retry up to 3 times in case of OpenAI server error.
-                try:
-                    transcript = st.session_state.openai_client.audio.transcriptions.create(
-                        model="whisper-1",
-                        file=audio_bio,
-                        language=language
-                    )
-                except Exception as e:
-                    print(str(e))  # log the exception in the terminal
-                    err += 1
-                else:
-                    success = True
-                    output = transcript.text
-                    st.session_state._last_speech_to_text_transcript = output
-        elif not just_once:
-            output = st.session_state._last_speech_to_text_transcript
+        if openai_api_key:
+            if not 'openai_client' in st.session_state:
+                #assert openai_api_key, openai_api_key
+                st.session_state.openai_client = OpenAI(api_key=openai_api_key)
+
+            id = audio['id']
+            new_output = (id > st.session_state._last_speech_to_text_transcript_id)
+            if new_output:
+                output = None
+                st.session_state._last_speech_to_text_transcript_id = id
+                audio_bio = io.BytesIO(audio['bytes'])
+                audio_bio.name = 'audio.webm'
+                success = False
+                err = 0
+                while not success and err < 3:  # Retry up to 3 times in case of OpenAI server error.
+                    try:
+                        transcript = st.session_state.openai_client.audio.transcriptions.create(
+                            model="whisper-1",
+                            file=audio_bio,
+                            language=language
+                        )
+                    except Exception as e:
+                        print(str(e))  # log the exception in the terminal
+                        err += 1
+                    else:
+                        success = True
+                        output = transcript.text
+                        st.session_state._last_speech_to_text_transcript = output
+            elif not just_once:
+                output = st.session_state._last_speech_to_text_transcript
+            else:
+                output = None
         else:
             output = None
-
     if key:
         st.session_state[key + '_output'] = output
     if new_output and callback: