Update openai_rag_chatbot.py

openai_rag_chatbot.py

@@ -6,7 +6,6 @@ import re
 import warnings
 import logging
 import streamlit as st
-from dotenv import load_dotenv
 from langdetect import detect
 
 from langchain_community.chat_models import ChatOpenAI
@@ -17,8 +16,9 @@ from langchain.indexes import VectorstoreIndexCreator
 from langchain.chains import RetrievalQA
 from langchain_core.prompts import ChatPromptTemplate
 
-# Load key
-load_dotenv()
+# 🔐 Embed your API key directly for Streamlit Cloud deployment
+OPENAI_API_KEY = "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"  # Replace with your actual key
+
 warnings.filterwarnings("ignore")
 logging.getLogger("transformers").setLevel(logging.ERROR)
 
@@ -42,7 +42,7 @@ def detect_language(text):
 
 @st.cache_resource
 def get_vectorstore():
-    pdf_path = "bad_breisig_docs.pdf"  # Adjust this if necessary
+    pdf_path = "bad_breisig_docs.pdf"  # Ensure this path is correct on Streamlit Cloud
     loaders = [PyPDFLoader(pdf_path)]
     return VectorstoreIndexCreator(
         embedding=HuggingFaceEmbeddings(model_name='all-MiniLM-L12-v2'),
@@ -52,8 +52,8 @@ def get_vectorstore():
     ).from_loaders(loaders).vectorstore
 
 def extract_political_groups(text):
-    pattern = re.compile(r'•\s.*(Gruppe|CDU|SPD|FDP|Union|Junge Union|AsF)', re.IGNORECASE)
-    return '\n'.join([m.group(0).strip() for m in pattern.finditer(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')
 
@@ -65,7 +65,7 @@ if prompt:
         openai_chat = ChatOpenAI(
             model_name="gpt-3.5-turbo",
             temperature=0,
-            openai_api_key=os.getenv("OPENAI_API_KEY")
+            openai_api_key=OPENAI_API_KEY
         )
 
         lang = detect_language(prompt)
@@ -101,7 +101,6 @@ Context:
             result = chain({"query": prompt})
             response = result["result"].strip()
 
-            # Special case: Political group list
             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)
@@ -109,7 +108,6 @@ Context:
                 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}"
 
-            # Fallback if no relevant info
             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})
@@ -124,4 +122,4 @@ Context:
         st.session_state.messages.append({'role': 'assistant', 'content': response})
 
     except Exception as e:
-        st.error(f"❌ Error: {str(e)}")
+        st.error(f"❌ Error: {str(e)}")