Spaces:
Sleeping
Sleeping
File size: 5,300 Bytes
f6697f7 e9dd7cd f6697f7 e5e8c03 f6697f7 e9dd7cd f6697f7 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 |
import os
os.environ["TOKENIZERS_PARALLELISM"] = "false"
os.environ["STREAMLIT_WATCHER_TYPE"] = "none"
import re
import warnings
import logging
import streamlit as st
from langdetect import detect
from langchain_community.chat_models import ChatOpenAI
from langchain_community.embeddings import HuggingFaceEmbeddings
from langchain.text_splitter import RecursiveCharacterTextSplitter
from langchain_community.document_loaders import PyPDFLoader
from langchain.indexes import VectorstoreIndexCreator
from langchain.chains import RetrievalQA
from langchain_core.prompts import ChatPromptTemplate
# 🔐 Embed your API key directly for Streamlit Cloud deployment
OPENAI_API_KEY = "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"
warnings.filterwarnings("ignore")
logging.getLogger("transformers").setLevel(logging.ERROR)
st.set_page_config(page_title="RegioWizard KI", layout="centered")
st.title('🧠 RegioWizard KI')
if 'messages' not in st.session_state:
st.session_state.messages = []
for message in st.session_state.messages:
st.chat_message(message['role']).markdown(message['content'])
def is_greeting(text):
return text.lower().strip() in ["hi", "hello", "hey", "greetings", "hallo", "servus", "moin"]
def detect_language(text):
try:
return detect(text)
except:
return "en"
@st.cache_resource
def get_vectorstore():
pdf_path = "Pavanmanagoli/regiowizard" # Ensure this path is correct on Streamlit Cloud
loaders = [PyPDFLoader(pdf_path)]
return VectorstoreIndexCreator(
embedding=HuggingFaceEmbeddings(model_name='all-MiniLM-L12-v2'),
text_splitter=RecursiveCharacterTextSplitter(
chunk_size=600, chunk_overlap=300, separators=["\n\n", "\n", ".", "•"]
)
).from_loaders(loaders).vectorstore
def extract_political_groups(text):
pattern = re.compile(r'(AsF|CDU|SPD|FDP|Junge Union|Senioren-Union|Freie W[aä]hlergruppe)[^\n]*', re.IGNORECASE)
return '\n'.join(sorted(set([m.group(0).strip() for m in pattern.finditer(text)])))
prompt = st.chat_input('Pass your prompt here')
if prompt:
st.chat_message('user').markdown(prompt)
st.session_state.messages.append({'role': 'user', 'content': prompt})
try:
openai_chat = ChatOpenAI(
model_name="gpt-3.5-turbo",
temperature=0,
openai_api_key=OPENAI_API_KEY
)
lang = detect_language(prompt)
if is_greeting(prompt):
response = "Hallo, ich bin der RegioWizard_KI Chatbot! 😊 Frag mich alles über Bad Breisig!" if lang == "de" else "Hi, I'm RegioWizard_KI Chatbot! 😊 Ask me anything about Bad Breisig!"
else:
vectorstore = get_vectorstore()
qa_prompt = ChatPromptTemplate.from_template("""
{prefix}
Context:
{context}
{q_prefix}: {question}
{a_prefix}:
""").partial(
prefix="Du bist ein hilfsbereiter Assistent mit Wissen über Bad Breisig. Verwende AUSSCHLIESSLICH den untenstehenden Kontext, um die Frage des Nutzers zu beantworten." if lang == "de" else "You are a helpful assistant knowledgeable about Bad Breisig. Use ONLY the context below to answer the user's question.",
q_prefix="Frage" if lang == "de" else "Question",
a_prefix="Antwort" if lang == "de" else "Answer"
)
chain = RetrievalQA.from_chain_type(
llm=openai_chat,
chain_type='stuff',
retriever=vectorstore.as_retriever(search_kwargs={'k': 12}),
chain_type_kwargs={"prompt": qa_prompt},
return_source_documents=True
)
result = chain({"query": prompt})
response = result["result"].strip()
if any(x in prompt.lower() for x in ["partei", "gruppierung", "gruppen", "parties", "political"]):
fallback_docs = result.get("source_documents", [])
combined_text = "\n".join(doc.page_content for doc in fallback_docs)
filtered = extract_political_groups(combined_text)
if filtered:
response = f"Die politischen Gruppierungen in Bad Breisig sind:\n\n{filtered}" if lang == "de" else f"The political groups in Bad Breisig are:\n\n{filtered}"
if not response or "not found" in response.lower() or "nicht im kontext" in response.lower():
fallback_docs = vectorstore.similarity_search_with_score(prompt, k=3)
keyword_hits = list({doc.page_content.strip()[:300] for doc, _ in fallback_docs})
if keyword_hits:
response = "Hier sind die relevantesten Informationen:\n\n" if lang == "de" else "Here’s the most relevant information found:\n\n"
response += "\n\n".join(keyword_hits)
else:
response = "Nicht im bereitgestellten Dokument gefunden." if lang == "de" else "Not found in the provided document."
st.chat_message('assistant').markdown(response)
st.session_state.messages.append({'role': 'assistant', 'content': response})
except Exception as e:
st.error(f"❌ Error: {str(e)}")
|