Merge pull request #196 from sudoleg/develop

minor improvements

Author: sudoleg

README.md

@@ -10,12 +10,11 @@ YouTubeGPT lets you **summarize and chat (Q&A)** with YouTube videos. Its featur
 
 - **provide a custom prompt for summaries** :writing_hand: [**VIEW DEMO**](https://youtu.be/rJqx3qvebws)
   - you can tailor the summary to your needs by providing a custom prompt or just use the default summarization
-- **automatically save summaries** :open_file_folder:
-  - the summaries can be automatically saved in the directory where you run the app. The summaries will be available under `<YT-channel-name>/<video-title>.md`
 - **get answers to questions about the video content** :question: [**VIEW DEMO**](https://youtu.be/rI8NogvHplE)
   - part of the application is designed and optimized specifically for question answering tasks (Q&A)
-- **create your own knowledge base**  :floppy_disk:
-  - once a video is processed, you can chat with it at any time!
+- **create your own library/knowledge base** :open_file_folder:
+  - the summaries and answers can be saved to a library accessible at a separate page!
+  - additionally, summaries can be automatically saved in the directory where you run the app. The summaries will be available under `<YT-channel-name>/<video-title>.md`
 - **choose from different OpenAI models** :robot:
   - currently available: gpt-3.5-turbo, gpt-4 (turbo), gpt-4o (mini)
   - by choosing a different model, you can summarize even longer videos and potentially get better responses

modules/persistance.py

@@ -117,5 +117,9 @@ def save_library_entry(
 
 def delete_library_entry(lib_entry: LibraryEntry):
     """Deletes a library entry."""
-    LibraryEntry.delete_by_id(lib_entry)
-    logging.info("Deleted library entry for video '%s'", lib_entry.video.title)
+    try:
+        LibraryEntry.delete_by_id(lib_entry)
+    except Exception as e:
+        logging.error("An error occured during the deletion of a library entry: %s", e)
+    else:
+        logging.info("Deleted library entry for video '%s'", lib_entry.video.title)

modules/youtube.py

@@ -13,7 +13,6 @@
 from .helpers import (
     extract_youtube_video_id,
     get_preffered_languages,
-    save_response_as_file,
 )
 
 OEMBED_PROVIDER = "https://noembed.com/embed"
@@ -56,12 +55,6 @@ def get_video_metadata(url: str):
         return None
     else:
         json_response = json.loads(response.text)
-        save_response_as_file(
-            dir_name="./video_meta",
-            filename=f"{json_response['title']}",
-            file_content=json_response,
-            content_type="json",
-        )
         return {
             "name": json_response["title"],
             "channel": json_response["author_name"],

pages/chat.py

@@ -17,7 +17,6 @@
     is_environment_prod,
     num_tokens_from_string,
     read_file,
-    save_response_as_file,
 )
 from modules.persistance import (
     SQL_DB,
@@ -286,11 +285,6 @@ def save_response_to_lib():
                             download_folder_path="data/audio",
                         )
                         whisper_transcript = generate_transcript(file_path=download)
-                        save_response_as_file(
-                            dir_name="data/transcripts",
-                            filename=saved_video.title,
-                            file_content=whisper_transcript,
-                        )
                         transcript_excerpts = split_text_recursively(
                             transcript_text=whisper_transcript,
                             chunk_size=chunk_size,

pages/library.py

@@ -1,6 +1,6 @@
 import streamlit as st
 
-from modules.persistance import SQL_DB, LibraryEntry, delete_library_entry
+from modules.persistance import SQL_DB, LibraryEntry, delete_library_entry, Video
 from modules.ui import display_nav_menu
 
 
@@ -13,12 +13,8 @@
 SQL_DB.create_tables([LibraryEntry], safe=True)
 # --- end ---
 
-saved_lib_entries_summaries = (
-    LibraryEntry.select().where(LibraryEntry.entry_type == "S").execute()
-)
-saved_lib_entries_answers = (
-    LibraryEntry.select().where(LibraryEntry.entry_type == "A").execute()
-)
+tab_summaries, tab_answers = st.tabs(["Summaries", "Answers"])
+saved_videos = Video.select()
 
 
 def execute_entry_deletion(entry: LibraryEntry):
@@ -27,32 +23,81 @@ def execute_entry_deletion(entry: LibraryEntry):
     st.rerun()
 
 
-if saved_lib_entries_summaries:
-    st.header("Saved summaries")
-    for i, entry in enumerate(saved_lib_entries_summaries, 0):
-        st.caption(f"{entry.video.title} - {entry.video.channel}")
-        with st.expander("Show"):
-            st.write(entry.text)
-        if st.button(
-            label="Delete entry",
-            key=f"delete_summary_{i}",
-        ):
-            execute_entry_deletion(entry)
-        st.divider()
-else:
-    st.info("You don't have any saved summaries yet!")
-
-if saved_lib_entries_answers:
-    st.header("Saved answers")
-    for j, entry in enumerate(saved_lib_entries_answers, 0):
-        st.subheader(entry.question)
-        with st.expander("Show"):
-            st.write(entry.text)
-        if st.button(
-            label="Delete entry",
-            key=f"delete_answer_{j}",
-        ):
-            execute_entry_deletion(entry)
-        st.divider()
-else:
-    st.info("You don't have any saved answers yet!")
+with tab_summaries:
+    selected_channel = st.selectbox(
+        label="Filter by channel",
+        placeholder="choose a channel or start typing",
+        # only videos with an associated transcript can be selected
+        options={video.channel for video in saved_videos},
+        index=None,
+        key="selected_channel",
+    )
+
+    if selected_channel:
+        saved_lib_entries_summaries = (
+            LibraryEntry.select()
+            .join(Video)
+            .where(LibraryEntry.entry_type == "S", Video.channel == selected_channel)
+            .execute()
+        )
+    else:
+        saved_lib_entries_summaries = (
+            LibraryEntry.select().where(LibraryEntry.entry_type == "S").execute()
+        )
+
+    if saved_lib_entries_summaries:
+        st.header("Saved summaries")
+        for i, entry in enumerate(saved_lib_entries_summaries, 0):
+            st.caption(f"{entry.video.title} - {entry.video.channel}")
+            with st.expander("Show"):
+                st.write(entry.text)
+            if st.button(
+                label="Delete entry",
+                key=f"delete_summary_{i}",
+            ):
+                execute_entry_deletion(entry)
+            st.divider()
+    else:
+        st.info("You don't have any saved summaries yet!")
+
+with tab_answers:
+    selected_video_title = st.selectbox(
+        label="Filter by video",
+        placeholder="choose a video or start typing",
+        # only videos with an associated transcript can be selected
+        options=[
+            video.title for video in saved_videos if video.transcripts.count() != 0
+        ],
+        index=None,
+        key="selected_video",
+    )
+
+    if selected_video_title:
+        saved_lib_entries_answers = (
+            LibraryEntry.select()
+            .join(Video)
+            .where(
+                LibraryEntry.entry_type == "A",
+                Video.title == st.session_state.selected_video,
+            )
+            .execute()
+        )
+    else:
+        saved_lib_entries_answers = (
+            LibraryEntry.select().where(LibraryEntry.entry_type == "A").execute()
+        )
+
+    if saved_lib_entries_answers:
+        st.header("Saved answers")
+        for j, entry in enumerate(saved_lib_entries_answers, 0):
+            st.subheader(entry.question)
+            with st.expander("Show"):
+                st.write(entry.text)
+            if st.button(
+                label="Delete entry",
+                key=f"delete_answer_{j}",
+            ):
+                execute_entry_deletion(entry)
+            st.divider()
+    else:
+        st.info("You don't have any saved answers yet!")

pages/summary.py

@@ -45,13 +45,22 @@
 display_link_to_repo("summary")
 # --- end ---
 
-# variable for holding the Video object
-saved_video: None | Video = None
+
+@st.dialog(title="Transcript too long", width="large")
+def display_dialog(message: str):
+    st.warning(message)
 
 
 def save_summary_to_lib():
     """Wrapper func for saving summaries to the library."""
     try:
+        saved_video = Video.create(
+            yt_video_id=extract_youtube_video_id(url_input),
+            link=url_input,
+            title=vid_metadata["name"],
+            channel=vid_metadata["channel"],
+            saved_on=dt.now(),
+        )
         save_library_entry(
             entry_type="S",
             question_text=None,
@@ -65,12 +74,7 @@ def save_summary_to_lib():
         st.success("Saved summary to library successfully!")
 
 
-@st.dialog(title="Transcript too long", width="large")
-def display_dialog(message: str):
-    st.warning(message)
-
-
-if is_api_key_set and is_api_key_valid(st.session_state.openai_api_key):
+if is_api_key_set() and is_api_key_valid(st.session_state.openai_api_key):
 
     # --- rest of the sidebar, which requires an api key to be set ---
     display_model_settings_sidebar()
@@ -113,13 +117,6 @@ def display_dialog(message: str):
     with col2:
         if summarize_button:
             try:
-                saved_video = Video.create(
-                    yt_video_id=extract_youtube_video_id(url_input),
-                    link=url_input,
-                    title=vid_metadata["name"],
-                    channel=vid_metadata["channel"],
-                    saved_on=dt.now(),
-                )
                 transcript = fetch_youtube_transcript(url_input)
                 cb = OpenAICallbackHandler()
                 llm = ChatOpenAI(
@@ -133,10 +130,14 @@ def display_dialog(message: str):
                 with st.spinner("Summarizing video :gear: Hang on..."):
                     if custom_prompt:
                         resp = get_transcript_summary(
-                            transcript, llm, custom_prompt=custom_prompt
+                            transcript_text=transcript,
+                            llm=llm,
+                            custom_prompt=custom_prompt,
                         )
                     else:
-                        resp = get_transcript_summary(transcript, llm)
+                        resp = get_transcript_summary(
+                            transcript_text=transcript, llm=llm
+                        )
                     st.session_state.summary = resp
                 st.markdown(st.session_state.summary)