app.py

# import nest_asyncio
# from youtube_transcript_api import YouTubeTranscriptApi
# import streamlit as st
# import os
# from groq import Groq

# nest_asyncio.apply()

# # --- CONFIGURATION ---
# YOUTUBE_API_KEY = os.environ.get("YOUTUBE_API_KEY")  # Set in your HuggingFace Secrets
# channel_id = "UCsv3kmQ5k1eIRG2R9mWN"  # @icodeguru0

# # Initialize Groq client once
# groq_client = Groq(api_key=os.environ.get("GROQ_API_KEY"))

# # --- FUNCTION: Fetch recent video IDs from YouTube channel ---
# def get_latest_video_ids(channel_id, max_results=5):
#     import requests
#     url = f"https://www.googleapis.com/youtube/v3/search?key={YOUTUBE_API_KEY}&channelId={channel_id}&part=snippet,id&order=date&maxResults={max_results}"
#     response = requests.get(url)
#     videos = response.json().get('items', [])
#     return [v['id']['videoId'] for v in videos if v['id']['kind'] == 'youtube#video']

# # --- FUNCTION: Get video transcripts ---
# def get_video_transcripts(video_ids):
#     all_transcripts = []
#     for vid in video_ids:
#         try:
#             transcript = YouTubeTranscriptApi.get_transcript(vid)
#             text = " ".join([t['text'] for t in transcript])
#             all_transcripts.append(text)
#         except:
#             continue
#     return all_transcripts

# # --- FUNCTION: Ask Groq API using official client ---
# def ask_groq(context, question):
#     messages = [
#         {"role": "system", "content": "You are a helpful assistant."},
#         {"role": "user", "content": f"Context: {context}\n\nQuestion: {question}\nAnswer:"}
#     ]
#     chat_completion = groq_client.chat.completions.create(
#         model="llama-3.3-70b-versatile",  # Or the model you have access to
#         messages=messages,
#     )
#     return chat_completion.choices[0].message.content.strip()

# # --- STREAMLIT APP ---
# def main():
#     st.set_page_config(page_title="EduBot - YouTube Channel QA", layout="wide")
#     st.title("🎓 EduBot for @icodeguru0")
#     st.markdown("Ask anything based on the channel’s recent videos.")

#     question = st.text_input("💬 Ask your question here:")
#     if question:
#         with st.spinner("🔍 Fetching videos and transcripts..."):
#             video_ids = get_latest_video_ids(channel_id)
#             transcripts = get_video_transcripts(video_ids)
#             full_context = "\n\n".join(transcripts)
#         with st.spinner("🧠 Thinking..."):
#             answer = ask_groq(full_context, question)
#         st.success(answer)

#     st.markdown("---")
#     st.caption("Powered by YouTube + Groq | Built for @icodeguru0")

# if __name__ == "__main__":
#     main()


# # import nest_asyncio
# # from youtube_transcript_api import YouTubeTranscriptApi
# # import streamlit as st
# # import os
# # from groq import Groq
# # import requests
# # from bs4 import BeautifulSoup

# # nest_asyncio.apply()

# # # --- CONFIGURATION ---
# # YOUTUBE_API_KEY = os.environ.get("YOUTUBE_API_KEY")  # Set in your HuggingFace Secrets
# # channel_id = "UCsv3kmQ5k1eIRG2R9mWN"  # @icodeguru0
# # BASE_URL = "https://icode.guru"

# # # Initialize Groq client once
# # groq_client = Groq(api_key=os.environ.get("GROQ_API_KEY"))

# # # --- FUNCTION: Fetch recent video IDs from YouTube channel ---
# # def get_latest_video_ids(channel_id, max_results=5):
# #     url = f"https://www.googleapis.com/youtube/v3/search?key={YOUTUBE_API_KEY}&channelId={channel_id}&part=snippet,id&order=date&maxResults={max_results}"
# #     response = requests.get(url)
# #     videos = response.json().get('items', [])
# #     return [v['id']['videoId'] for v in videos if v['id']['kind'] == 'youtube#video']

# # # --- FUNCTION: Get video transcripts ---
# # def get_video_transcripts(video_ids):
# #     all_transcripts = []
# #     for vid in video_ids:
# #         try:
# #             transcript = YouTubeTranscriptApi.get_transcript(vid)
# #             text = " ".join([t['text'] for t in transcript])
# #             all_transcripts.append(text)
# #         except:
# #             continue
# #     return all_transcripts

# # # --- NEW FUNCTION: Scrape icode.guru ---
# # def scrape_icodeguru(base_url="https://icode.guru", max_pages=5):
# #     visited = set()
# #     blocks = []

# #     def crawl(url):
# #         if url in visited or len(visited) >= max_pages:
# #             return
# #         visited.add(url)
# #         try:
# #             res = requests.get(url, timeout=10)
# #             soup = BeautifulSoup(res.content, "html.parser")
# #             page_text = soup.get_text(separator=" ", strip=True)
# #             if len(page_text) > 100:
# #                 blocks.append(f"[Source]({url}):\n{page_text[:2000]}")
# #             for link in soup.find_all("a", href=True):
# #                 href = link['href']
# #                 if href.startswith("/"):
# #                     href = base_url + href
# #                 if href.startswith(base_url):
# #                     crawl(href)
# #         except:
# #             pass

# #     crawl(base_url)
# #     return blocks

# # # --- FUNCTION: Ask Groq API using official client ---
# # def ask_groq(context, question):
# #     messages = [
# #         {"role": "system", "content": "You are a helpful assistant. Only answer using the given context (YouTube + icode.guru). Provide links if possible."},
# #         {"role": "user", "content": f"Context:\n{context}\n\nQuestion: {question}\nAnswer:"}
# #     ]
# #     chat_completion = groq_client.chat.completions.create(
# #         model="llama-3.3-70b-versatile",  # Or the model you have access to
# #         messages=messages,
# #     )
# #     return chat_completion.choices[0].message.content.strip()

# # # --- STREAMLIT APP ---
# # def main():
# #     st.set_page_config(page_title="EduBot - YouTube + iCodeGuru QA", layout="wide")
# #     st.title("🎓 EduBot for @icodeguru0")
# #     st.markdown("Ask anything based on the channel’s recent videos and website content from [icode.guru](https://icode.guru).")

# #     question = st.text_input("💬 Ask your question here:")
# #     if question:
# #         with st.spinner("🔍 Fetching videos and transcripts..."):
# #             video_ids = get_latest_video_ids(channel_id)
# #             transcripts = get_video_transcripts(video_ids)
# #             yt_context = "\n\n".join(transcripts)

# #         with st.spinner("🌐 Scraping icode.guru..."):
# #             site_blocks = scrape_icodeguru(BASE_URL, max_pages=5)
# #             site_context = "\n\n".join(site_blocks)

# #         full_context = yt_context + "\n\n" + site_context

# #         with st.spinner("🧠 Thinking..."):
# #             answer = ask_groq(full_context, question)
# #         st.success(answer)

# #     st.markdown("---")
# #     st.caption("Powered by YouTube + iCodeGuru + Groq | Built for @icodeguru0")

# # if __name__ == "__main__":
# #     main()


#(youtube+web)


# import nest_asyncio
# import streamlit as st
# import os
# import requests
# from youtube_transcript_api import YouTubeTranscriptApi
# from groq import Groq
# from bs4 import BeautifulSoup

# nest_asyncio.apply()

# # --- CONFIGURATION ---
# YOUTUBE_API_KEY = os.environ.get("YOUTUBE_API_KEY")
# GROQ_API_KEY = os.environ.get("GROQ_API_KEY")
# channel_id = "UCsv3kmQ5k1eIRG2R9mWN"  # iCodeGuru
# BASE_URL = "https://icode.guru"

# groq_client = Groq(api_key=GROQ_API_KEY)

# # --- Fetch recent video IDs from YouTube channel ---
# def get_latest_video_ids(channel_id, max_results=5):
#     url = f"https://www.googleapis.com/youtube/v3/search?key={YOUTUBE_API_KEY}&channelId={channel_id}&part=snippet,id&order=date&maxResults={max_results}"
#     response = requests.get(url)
#     videos = response.json().get('items', [])
    
#     valid_videos = []
#     for v in videos:
#         if v['id']['kind'] == 'youtube#video':
#             title = v['snippet']['title']
#             channel_title = v['snippet']['channelTitle']
#             video_id = v['id']['videoId']
#             if "icodeguru" in channel_title.lower():  # ✅ Extra validation
#                 valid_videos.append((video_id, title))
#     return valid_videos


# # --- Get video transcripts ---
# def get_video_transcripts(video_info):
#     results = []
#     for vid, title in video_info:
#         try:
#             transcript = YouTubeTranscriptApi.get_transcript(vid)
#             text = " ".join([t['text'] for t in transcript])
#             video_link = f"https://www.youtube.com/watch?v={vid}"
#             results.append({
#                 "video_id": vid,
#                 "title": title,
#                 "link": video_link,
#                 "transcript": text
#             })
#         except Exception as e:
#             continue
#     return results

# # --- Scrape icode.guru ---
# def scrape_icodeguru(base_url=BASE_URL, max_pages=5):
#     visited = set()
#     blocks = []

#     def crawl(url):
#         if url in visited or len(visited) >= max_pages:
#             return
#         visited.add(url)
#         try:
#             res = requests.get(url, timeout=10)
#             soup = BeautifulSoup(res.content, "html.parser")
#             page_text = soup.get_text(separator=" ", strip=True)
#             if len(page_text) > 100:
#                 blocks.append(f"[{url}]({url}):\n{page_text[:1500]}")
#             for link in soup.find_all("a", href=True):
#                 href = link['href']
#                 if href.startswith("/"):
#                     href = base_url + href
#                 if href.startswith(base_url):
#                     crawl(href)
#         except:
#             pass

#     crawl(base_url)
#     return blocks

# # --- Ask Groq ---
# def ask_groq(context, question):
#     messages = [
#         {"role": "system", "content": "You are a helpful assistant. Always provide relevant video and website links if possible."},
#         {"role": "user", "content": f"Context:\n{context}\n\nQuestion: {question}\nAnswer (include links):"}
#     ]
#     chat_completion = groq_client.chat.completions.create(
#         model="llama-3.3-70b-versatile",
#         messages=messages,
#     )
#     return chat_completion.choices[0].message.content.strip()

# #--- STREAMLIT APP ---
# def main():
#     st.set_page_config(page_title="EduBot for iCodeGuru", layout="wide")
#     st.title("🎓 EduBot for @icodeguru0")
#     st.markdown("Ask anything based on the latest YouTube videos and website content of [icode.guru](https://icode.guru).")

#     question = st.text_input("💬 Ask your question:")
#     if question:
#         with st.spinner("📺 Fetching YouTube videos..."):
#             video_info = get_latest_video_ids(channel_id, max_results=5)
#             transcripts = get_video_transcripts(video_info)

#         yt_context = ""
#         relevant_links = []
#         for vid in transcripts:
#             yt_context += f"\n\n[Video: {vid['title']}]({vid['link']}):\n{vid['transcript'][:1500]}"
#             if question.lower() in vid['transcript'].lower():
#                 relevant_links.append(vid['link'])

#         with st.spinner("🌐 Scraping icode.guru..."):
#             site_blocks = scrape_icodeguru(BASE_URL, max_pages=5)
#             site_context = "\n\n".join(site_blocks)

#         full_context = yt_context + "\n\n" + site_context

#         with st.spinner("🧠 Thinking..."):
#             answer = ask_groq(full_context, question)

#         st.success(answer)

#         if relevant_links:
#             st.markdown("### 🔗 Related YouTube Links")
#             for link in relevant_links:
#                 st.markdown(f"- [Watch Video]({link})")

#     st.markdown("---")
#     st.caption("Powered by YouTube, iCodeGuru, and Groq")

# if __name__ == "__main__":
#     main()

# (vectordb)


# import nest_asyncio
# import streamlit as st
# import os
# import requests
# from youtube_transcript_api import YouTubeTranscriptApi
# from groq import Groq
# from bs4 import BeautifulSoup
# from sentence_transformers import SentenceTransformer
# import chromadb
# from chromadb.utils.embedding_functions import SentenceTransformerEmbeddingFunction

# import json

# nest_asyncio.apply()

# # --- CONFIGURATION ---
# YOUTUBE_API_KEY = os.environ.get("YOUTUBE_API_KEY")
# GROQ_API_KEY = os.environ.get("GROQ_API_KEY")
# channel_id = "UCsv3kmQ5k1eIRG2R9mWN"  # iCodeGuru
# BASE_URL = "https://icode.guru"

# groq_client = Groq(api_key=GROQ_API_KEY)
# from chromadb.utils.embedding_functions import SentenceTransformerEmbeddingFunction

# embedding_function = SentenceTransformerEmbeddingFunction("all-MiniLM-L6-v2")

# chroma_client = chromadb.Client()
# collection = chroma_client.get_or_create_collection("icodeguru_knowledge", embedding_function=embedding_function)

# # --- Upload + load files as vector DB ---
# def load_uploaded_vectors(uploaded_files):
#     data = []
#     for file in uploaded_files:
#         if file.name.endswith(".txt"):
#             text = file.read().decode()
#             data.append({"id": file.name, "content": text})
#         elif file.name.endswith(".json"):
#             content = json.load(file)
#             for i, chunk in enumerate(content):
#                 data.append({"id": f"{file.name}-{i}", "content": chunk})
#     return data

# def search_vector_data(query, data):
#     if not data:
#         return None
#     collection = chroma_client.get_or_create_collection("temp_query", embedding_function=embedding_function)
#     collection.add(documents=[d["content"] for d in data], ids=[d["id"] for d in data])
#     results = collection.query(query_texts=[query], n_results=3)
#     if results and results["documents"]:
#         return "\n\n".join([doc for doc in results["documents"][0]])
#     return None

# # --- Fetch recent video IDs from YouTube channel ---
# def get_latest_video_ids(channel_id, max_results=5):
#     url = f"https://www.googleapis.com/youtube/v3/search?key={YOUTUBE_API_KEY}&channelId={channel_id}&part=snippet,id&order=date&maxResults={max_results}"
#     response = requests.get(url)
#     videos = response.json().get('items', [])
    
#     valid_videos = []
#     for v in videos:
#         if v['id']['kind'] == 'youtube#video':
#             title = v['snippet']['title']
#             channel_title = v['snippet']['channelTitle']
#             video_id = v['id']['videoId']
#             if "icodeguru" in channel_title.lower():
#                 valid_videos.append((video_id, title))
#     return valid_videos

# # --- Get video transcripts ---
# def get_video_transcripts(video_info):
#     results = []
#     for vid, title in video_info:
#         try:
#             transcript = YouTubeTranscriptApi.get_transcript(vid)
#             text = " ".join([t['text'] for t in transcript])
#             video_link = f"https://www.youtube.com/watch?v={vid}"
#             results.append({
#                 "video_id": vid,
#                 "title": title,
#                 "link": video_link,
#                 "transcript": text
#             })
#         except:
#             continue
#     return results

# # --- Scrape icode.guru ---
# def scrape_icodeguru(base_url=BASE_URL, max_pages=5):
#     visited = set()
#     blocks = []

#     def crawl(url):
#         if url in visited or len(visited) >= max_pages:
#             return
#         visited.add(url)
#         try:
#             res = requests.get(url, timeout=10)
#             soup = BeautifulSoup(res.content, "html.parser")
#             page_text = soup.get_text(separator=" ", strip=True)
#             if len(page_text) > 100:
#                 blocks.append(f"[{url}]({url}):\n{page_text[:1500]}")
#             for link in soup.find_all("a", href=True):
#                 href = link['href']
#                 if href.startswith("/"):
#                     href = base_url + href
#                 if href.startswith(base_url):
#                     crawl(href)
#         except:
#             pass

#     crawl(base_url)
#     return blocks

# # --- Ask Groq ---
# def ask_groq(context, question):
#     messages = [
#         {"role": "system", "content": "You are a helpful assistant. Always provide relevant video and website links if possible."},
#         {"role": "user", "content": f"Context:\n{context}\n\nQuestion: {question}\nAnswer (include links):"}
#     ]
#     chat_completion = groq_client.chat.completions.create(
#         model="llama3-8b-8192",
#         messages=messages,
#     )
#     return chat_completion.choices[0].message.content.strip()

# #--- STREAMLIT APP ---
# def main():
#     st.set_page_config(page_title="EduBot for iCodeGuru", layout="wide")
#     st.title("🎓 EduBot for @icodeguru0")
#     st.markdown("Ask anything based on the latest YouTube videos and website content of [icode.guru](https://icode.guru).")

#     uploaded_files = st.file_uploader("📁 Optionally upload your knowledge files (txt or json)", type=['txt', 'json'], accept_multiple_files=True)
#     user_question = st.text_input("💬 Ask your question:")

#     if user_question:
#         vector_data = load_uploaded_vectors(uploaded_files) if uploaded_files else []

#         # Try vector DB first
#         vector_context = search_vector_data(user_question, vector_data)
#         if vector_context:
#             with st.spinner("🧠 Answering from uploaded knowledge..."):
#                 answer = ask_groq(vector_context, user_question)
#                 st.success(answer)
#         else:
#             # Fallback to real-time data
#             with st.spinner("📺 Fetching YouTube videos..."):
#                 video_info = get_latest_video_ids(channel_id, max_results=5)
#                 transcripts = get_video_transcripts(video_info)

#             yt_context = ""
#             relevant_links = []
#             for vid in transcripts:
#                 yt_context += f"\n\n[Video: {vid['title']}]({vid['link']}):\n{vid['transcript'][:1500]}"
#                 if user_question.lower() in vid['transcript'].lower():
#                     relevant_links.append(vid['link'])

#             with st.spinner("🌐 Scraping icode.guru..."):
#                 site_blocks = scrape_icodeguru(BASE_URL, max_pages=5)
#                 site_context = "\n\n".join(site_blocks)

#             full_context = yt_context + "\n\n" + site_context

#             with st.spinner("🧠 Thinking..."):
#                 answer = ask_groq(full_context, user_question)
#                 st.success(answer)

#             if relevant_links:
#                 st.markdown("### 🔗 Related YouTube Links")
#                 for link in relevant_links:
#                     st.markdown(f"- [Watch Video]({link})")

#     st.markdown("---")
#     st.caption("Powered by YouTube, iCodeGuru, and Groq")

# if __name__ == "__main__":
#     main()



# import nest_asyncio
# import streamlit as st
# import os
# from groq import Groq
# from sentence_transformers import SentenceTransformer
# import chromadb
# from chromadb.utils.embedding_functions import SentenceTransformerEmbeddingFunction

# nest_asyncio.apply()

# # --- CONFIGURATION ---
# GROQ_API_KEY = os.environ.get("GROQ_API_KEY")

# groq_client = Groq(api_key=GROQ_API_KEY)
# embedding_function = SentenceTransformerEmbeddingFunction(
#     model_name="all-MiniLM-L6-v2",
#     device="cpu"
# )


# chroma_client = chromadb.Client()
# collection = chroma_client.get_or_create_collection("icodeguru_knowledge", embedding_function=embedding_function)

# # --- Search persistent vector DB ---
# def search_vector_data(query):
#     results = collection.query(query_texts=[query], n_results=3)
#     if results and results["documents"]:
#         return "\n\n".join([doc for doc in results["documents"][0]])
#     return None

# # --- Ask Groq ---
# def ask_groq(context, question):
#     messages = [
#         {"role": "system", "content": "You are a helpful assistant. Always provide relevant video and website links if possible."},
#         {"role": "user", "content": f"Context:\n{context}\n\nQuestion: {question}\nAnswer (include links):"}
#     ]
#     chat_completion = groq_client.chat.completions.create(
#         model="llama3-8b-8192",
#         messages=messages,
#     )
#     return chat_completion.choices[0].message.content.strip()

# #--- STREAMLIT APP ---
# def main():
#     st.set_page_config(page_title="EduBot for iCodeGuru", layout="wide")
#     st.title("🎓 EduBot for @icodeguru0")
#     st.markdown("Ask anything based on pre-loaded iCodeGuru knowledge.")

#     user_question = st.text_input("💬 Ask your question:")

#     if user_question:
#         # Try vector DB first
#         vector_context = search_vector_data(user_question)
#         if vector_context:
#             with st.spinner("🧠 Answering from knowledge base..."):
#                 answer = ask_groq(vector_context, user_question)
#                 st.success(answer)
#         else:
#             st.warning("⚠️ No relevant answer found in the embedded knowledge.")

#     st.markdown("---")
#     st.caption("Powered by ChromaDB 🧠 and Groq ⚡")

# if __name__ == "__main__":
#     main()





# import nest_asyncio
# import streamlit as st
# import os
# from groq import Groq
# from sentence_transformers import SentenceTransformer
# import chromadb
# from chromadb.utils.embedding_functions import SentenceTransformerEmbeddingFunction
# from chromadb.config import Settings

# # Apply asyncio patch (required for nested event loops in Streamlit)
# nest_asyncio.apply()

# # --- CONFIGURATION ---
# GROQ_API_KEY = os.environ.get("GROQ_API_KEY")
# GROQ_MODEL = "llama3-8b-8192"

# # Initialize Groq client
# groq_client = Groq(api_key=GROQ_API_KEY)

# # Initialize embedding model
# embedding_function = SentenceTransformerEmbeddingFunction(
#     model_name="all-MiniLM-L6-v2",
#     device="cpu"
# )

# # Initialize ChromaDB with persistence
# chroma_client = chromadb.PersistentClient(path="./chroma_db", settings=Settings(anonymized_telemetry=False))
# collection = chroma_client.get_or_create_collection(
#     name="icodeguru_knowledge",
#     embedding_function=embedding_function
# )

# # --- Search embedded knowledge ---
# def search_vector_data(query):
#     try:
#         results = collection.query(query_texts=[query], n_results=3)
#         if results and results["documents"]:
#             return "\n\n".join(results["documents"][0])
#     except Exception as e:
#         st.error(f"Vector search error: {e}")
#     return None

# # --- Ask Groq ---
# def ask_groq(context, question):
#     messages = [
#         {"role": "system", "content": "You are a helpful assistant. Always provide relevant video and website links if possible."},
#         {"role": "user", "content": f"Context:\n{context}\n\nQuestion: {question}\nAnswer (include links):"}
#     ]
#     response = groq_client.chat.completions.create(
#         model=GROQ_MODEL,
#         messages=messages
#     )
#     return response.choices[0].message.content.strip()

# # --- Streamlit UI ---
# def main():
#     st.set_page_config(page_title="EduBot for iCodeGuru", layout="wide")
#     st.title("🎓 EduBot for @icodeguru0")
#     st.markdown("Ask anything based on pre-loaded iCodeGuru knowledge.")

#     user_question = st.text_input("💬 Ask your question:")

#     if user_question:
#         vector_context = search_vector_data(user_question)
#         if vector_context:
#             with st.spinner("🧠 Answering from knowledge base..."):
#                 answer = ask_groq(vector_context, user_question)
#                 st.success(answer)
#         else:
#             st.warning("⚠️ No relevant answer found in the embedded knowledge.")

#     st.markdown("---")
#     st.caption("Powered by ChromaDB 🧠 and Groq ⚡")


#     # ✅ This is the correct way to run the app
# if __name__ == "__main__":
#     main()







import nest_asyncio
import streamlit as st
import os
import json
from groq import Groq
from sentence_transformers import SentenceTransformer
import chromadb
from chromadb.utils.embedding_functions import SentenceTransformerEmbeddingFunction
from chromadb.config import Settings
from langchain.document_loaders import JSONLoader
from langchain.text_splitter import RecursiveCharacterTextSplitter

# Apply asyncio patch (Streamlit fix)
nest_asyncio.apply()

# --- CONFIGURATION ---
GROQ_API_KEY = os.environ.get("GROQ_API_KEY")
GROQ_MODEL = "llama3-8b-8192"

# Initialize Groq client
groq_client = Groq(api_key=GROQ_API_KEY)

# # Explicitly load SentenceTransformer model first to avoid meta tensor bug
# embedding_model = SentenceTransformer "(all-MiniLM-L6-v2)"

# # Pass this model into Chroma's embedding function
# embedding_function = SentenceTransformerEmbeddingFunction(embedding_model=embedding_model)

embedding_function = SentenceTransformerEmbeddingFunction(
    model_name="all-MiniLM-L6-v2",
    device="cpu"
)



# Initialize ChromaDB Persistent Client
chroma_client = chromadb.PersistentClient(path="./chroma_db", settings=Settings(anonymized_telemetry=False))
collection = chroma_client.get_or_create_collection(
    name="icodeguru_knowledge",
    embedding_function=embedding_function
)

# --- Ingest JSON Files from /docs/ ---
def ingest_docs_to_chroma():
    folder_path = "./docs"
    all_docs = []
    for filename in os.listdir(folder_path):
        if filename.endswith(".json"):
            file_path = os.path.join(folder_path, filename)
            loader = JSONLoader(file_path=file_path, jq_schema='.[]')
            docs = loader.load()
            all_docs.extend(docs)
            st.write(f"Loaded {len(docs)} documents from {filename}")

    # Chunk Documents
    text_splitter = RecursiveCharacterTextSplitter(chunk_size=500, chunk_overlap=50)
    chunks = text_splitter.split_documents(all_docs)
    st.write(f"Total chunks created: {len(chunks)}")

    # Add Chunks to ChromaDB
    for chunk in chunks:
        # Flatten list content if necessary
        if isinstance(chunk.page_content, list):
            content = " ".join(str(item) for item in chunk.page_content).strip()
        else:
            content = str(chunk.page_content).strip()
        
        metadata = chunk.metadata
        doc_id = str(hash(content))
        collection.add(documents=[content], metadatas=[metadata], ids=[doc_id])


    st.success("✅ Knowledge Base Updated Successfully!")

# --- Search embedded knowledge ---
def search_vector_data(query):
    try:
        results = collection.query(query_texts=[query], n_results=3)
        if results and results["documents"]:
            return "\n\n".join(results["documents"][0])
    except Exception as e:
        st.error(f"Vector search error: {e}")
    return None

# --- Ask Groq LLM ---
def ask_groq(context, question):
    messages = [
        {"role": "system", "content": "You are a helpful assistant. Always provide relevant video and website links if possible."},
        {"role": "user", "content": f"Context:\n{context}\n\nQuestion: {question}\nAnswer (include links):"}
    ]
    response = groq_client.chat.completions.create(
        model=GROQ_MODEL,
        messages=messages
    )
    return response.choices[0].message.content.strip()

# --- Streamlit UI ---
def main():
    st.set_page_config(page_title="EduBot for iCodeGuru", layout="wide")
    st.title("🎓 EduBot for @icodeguru0")
    st.markdown("Ask anything based on pre-loaded iCodeGuru knowledge.")

    # --- Auto Update Knowledge Base at App Start ---
    st.info("🔄 Updating Knowledge Base from /docs/...")
    ingest_docs_to_chroma()
    st.success("✅ Knowledge Base Loaded Successfully!")

    st.markdown("---")

    user_question = st.text_input("💬 Ask your question:")

    if user_question:
        vector_context = search_vector_data(user_question)
        if vector_context:
            with st.spinner("🧠 Answering from knowledge base..."):
                answer = ask_groq(vector_context, user_question)
                st.success(answer)
        else:
            st.warning("⚠️ No relevant answer found in the embedded knowledge.")

    st.markdown("---")
    st.caption("Powered by ChromaDB 🧠 and Groq ⚡")

if __name__ == "__main__":
    main()