feat: summarize page in UI

ui/main.py

@@ -25,6 +25,7 @@
         st.Page(page="pages/chat.py", title="Chat", icon=":material/chat:"),
         st.Page(page="pages/documents.py", title="Documents", icon=":material/file_copy:"),
         st.Page(page="pages/transcription.py", title="Transcription", icon=":material/graphic_eq:"),
+        st.Page(page="pages/summarize.py", title="Summarize", icon=":material/contract_edit:"),
     ]
 )
 pg.run()

ui/pages/chat.py

@@ -1,11 +1,11 @@
 import logging
 import traceback
 
-
 import streamlit as st
 
 from config import INTERNET_COLLECTION_DISPLAY_ID
-from utils import generate_stream, get_collections, get_models, header
+from utils.chat import generate_stream
+from utils.common import get_collections, get_models, header
 
 API_KEY = header()
 

ui/pages/documents.py

@@ -3,7 +3,8 @@
 import streamlit as st
 
 from config import PRIVATE_COLLECTION_TYPE
-from utils import create_collection, delete_collection, delete_document, header, load_data, upload_file
+from utils.common import header
+from utils.documents import create_collection, delete_collection, delete_document, load_data, upload_file
 
 API_KEY = header()
 

ui/pages/summarize.py

@@ -0,0 +1,175 @@
+import time
+
+from openai import OpenAI
+import streamlit as st
+from streamlit_extras.stylable_container import stylable_container
+
+from config import BASE_URL, PRIVATE_COLLECTION_TYPE
+from utils.common import get_collections, get_documents, get_models, header
+from utils.summarize import generate_toc
+
+# Config
+API_KEY = header()
+
+openai_client = OpenAI(base_url=BASE_URL, api_key=API_KEY)
+
+# Data
+try:
+    language_models, embeddings_models, _, _ = get_models(api_key=API_KEY)
+    collections = get_collections(api_key=API_KEY)
+    collections = [collection for collection in collections if collection["type"] == PRIVATE_COLLECTION_TYPE]
+    documents = get_documents(api_key=API_KEY, collection_ids=[collection["id"] for collection in collections])
+except Exception as e:
+    st.error(body="Error to fetch user data.")
+    st.stop()
+
+# State
+if "selected_model" not in st.session_state:
+    st.session_state.selected_model = language_models[0]
+
+# Sidebar
+with st.sidebar:
+    new_chat = st.button(label="**:material/refresh: New summarize**", key="new", use_container_width=True)
+    if new_chat:
+        st.session_state.pop("collection_id", None)
+        st.session_state.pop("document_id", None)
+        st.session_state.pop("validate_toc", None)
+        st.session_state.pop("toc", None)
+        st.session_state.pop("summary", None)
+        st.rerun()
+    params = {"sampling_params": dict()}
+
+    st.subheader(body="Summarize parameters")
+    st.session_state["selected_model"] = st.selectbox(
+        label="Language model", options=language_models, index=language_models.index(st.session_state.selected_model)
+    )
+    params["sampling_params"]["model"] = st.session_state["selected_model"]
+    params["sampling_params"]["temperature"] = st.slider(label="Temperature", value=0.2, min_value=0.0, max_value=1.0, step=0.1)
+
+# Main
+## Document
+st.session_state.collection_id = None if "collection_id" not in st.session_state else st.session_state.collection_id
+st.session_state.document_id = None if "document_id" not in st.session_state else st.session_state.document_id
+
+st.subheader(":material/counter_1: Select a document")
+st.info("Please select a document to generate a summary.")
+document = st.selectbox("Document", options=[f"{document["id"]} - {document["name"]}" for document in documents])
+document_id = document.split(" - ")[0]
+collection_id = [document["collection_id"] for document in documents if document["id"] == document_id][0]
+
+with stylable_container(
+    key="Summarize",
+    css_styles="""
+    button{
+        float: right;
+    }
+    """,
+):
+    if st.button(label="Validate", key="validate_document"):
+        st.session_state.collection_id = collection_id
+        st.session_state.document_id = document_id
+        st.toast("Document validated !", icon="✅")
+        time.sleep(2)
+        st.rerun()
+
+## TOC
+st.session_state.toc = "" if "toc" not in st.session_state else st.session_state.toc
+st.session_state.validate_toc = False if "validate_toc" not in st.session_state else st.session_state.validate_toc
+
+st.markdown("***")
+st.markdown(f"**state:**\n{st.session_state.toc}")
+st.subheader(":material/counter_2: Create a table of content")
+
+if not st.session_state.document_id:
+    st.stop()
+
+st.info(
+    body="For help the model to generate a summarize, you need to write a table of content of your document. Clic on *generate* button if you need an AI help."
+)
+
+toc = st.text_area(label="Table of content", value=st.session_state.toc, height=200)
+
+with stylable_container(
+    key="Summarize",
+    css_styles="""
+    .left-button {
+        float: left;
+    }
+    .right-button {
+        float: right;
+    .stButton {
+        width: auto;
+    }
+    """,
+):
+    col1, col2, col3 = st.columns([2, 6, 2])
+    with col1:
+        if st.button(label="✨ Generate", key="generate_toc", use_container_width=True):
+            with st.spinner("✨ Generate..."):
+                st.session_state.toc = generate_toc(
+                    collection_id=collections[0]["id"], document_id=document_id, api_key=API_KEY, model=st.session_state.selected_model
+                )
+                st.toast(body="Table of content generated !", icon="✅")
+                time.sleep(2)
+                st.rerun()
+    with col3:
+        if st.button(label="Validate", use_container_width=True):
+            if toc:
+                st.session_state.toc = toc
+                st.session_state.validate_toc = True
+                st.toast(body="Table of content validated !", icon="✅")
+                time.sleep(2)
+                st.rerun()
+            else:
+                st.toast(body="You have to write a table of content before validate it.", icon="❌")
+
+
+## Summary
+st.session_state.summary = "" if "summary" not in st.session_state else st.session_state.summary
+
+st.markdown(body="***")
+st.markdown(body=f"**state:**\n{st.session_state.summary}")
+st.subheader(body=":material/counter_3: Generate a summary")
+
+if not st.session_state.validate_toc:
+    st.stop()
+
+st.info("Here is the summary of your document.")
+
+st.write("**Summary**")
+st.code(st.session_state.summary, language="markdown", wrap_lines=True)
+
+with stylable_container(
+    key="Summarize",
+    css_styles="""
+    button{
+        float: right;
+    }
+    """,
+):
+    if st.button("Generate", key="generate_summary"):
+        with st.spinner("✨ Generate..."):
+            st.session_state.summary = "Generation"  # @TODO
+            st.toast("Summary generated !", icon="✅")
+        time.sleep(2)
+        st.rerun()
+
+feedback = st.text_area(label="Feedback")
+
+with stylable_container(
+    key="Summarize",
+    css_styles="""
+    button{
+        float: right;
+    }
+    """,
+):
+    if st.button("Give feedback", key="give_feedback"):
+        if st.session_state.summary:
+            with st.spinner("✨ Generate..."):
+                st.session_state.summary = "Generation with feedback"  # @TODO
+                st.toast("Summary generated with feedback !", icon="✅")
+                time.sleep(2)
+                st.rerun()
+        else:
+            st.toast("You have to generate a summary before give feedback.", icon="❌")

ui/pages/transcription.py

@@ -5,15 +5,15 @@
 import streamlit as st
 
 from config import BASE_URL, SUPPORTED_LANGUAGES
-from utils import get_models, header
+from utils.common import get_models, header
 
 API_KEY = header()
 
 # Data
 try:
     _, _, audio_models, _ = get_models(api_key=API_KEY)
 except Exception:
-    st.error("Error to fetch user data.")
+    st.error(body="Error to fetch user data.")
     logging.error(traceback.format_exc())
     st.stop()
 
@@ -22,28 +22,28 @@
 # Sidebar
 with st.sidebar:
     params = {}
-    st.subheader("Audio parameters")
+    st.subheader(body="Audio parameters")
     params["model"] = st.selectbox(label="Audio model", options=audio_models)
     params["temperature"] = st.slider(label="Temperature", value=0.2, min_value=0.0, max_value=1.0, step=0.1)
     params["language"] = st.selectbox(label="Language", options=SUPPORTED_LANGUAGES, index=SUPPORTED_LANGUAGES.index("french"))
 
 # Main
-col1, col2 = st.columns(2)
-file = st.file_uploader("Upload an audio file", type=["mp3", "wav", "m4a"])
+col1, col2 = st.columns(spec=2)
+file = st.file_uploader(label="Upload an audio file", type=["mp3", "wav", "m4a"])
 record = st.audio_input(label="Record a voice message")
 
 if file and record:
-    st.error("Please upload only one file at a time.")
+    st.error(body="Please upload only one file at a time.")
     st.stop()
 
 audio = record or file
 result = None
 _, center, _ = st.columns(spec=3)
 with center:
-    submit = st.button("**Transcribe**", use_container_width=True)
+    submit = st.button(label="**Transcribe**", use_container_width=True)
 
 if submit and audio:
-    with st.spinner("Transcribing audio..."):
+    with st.spinner(text="Transcribing audio..."):
         try:
             response = openai_client.audio.transcriptions.create(
                 file=audio,
@@ -52,9 +52,9 @@
             )
             result = response.text
         except Exception:
-            st.error("Error transcribing audio.")
+            st.error(body="Error transcribing audio.")
             logging.error(traceback.format_exc())
 
 if result:
-    st.caption("Result")
-    st.code(result, language="markdown", wrap_lines=True)
+    st.caption(body="Result")
+    st.code(body=result, language="markdown", wrap_lines=True)