Upload app.py

app.py

@@ -0,0 +1,117 @@
+# Load the packages and configurations
+import torch
+import pandas as pd
+from retriever import BM25Plus
+import os
+import streamlit as st
+from groq import Groq
+
+# Configure GROQ API_KEY
+client = Groq(
+    api_key="GROQ_API_KEY",
+)
+
+# RTL formatting of the streamlit
+st.markdown("""
+<style>
+body, html {
+    direction: RTL;
+    unicode-bidi: plaintext;
+    text-align: right;
+}
+p, div, input, label, h1, h2, h3, h4, h5, h6 {
+    direction: RTL;
+    unicode-bidi: plaintext;
+    text-align: right;
+}
+</style>
+""", unsafe_allow_html=True)
+
+# Streamlit details
+st.title("DarooGAP")
+st.sidebar.title("💊 داروگپ 💊")
+st.sidebar.divider()
+
+# Load the dataset
+df = pd.read_csv('darooyab_qa.csv')
+corpus = df.loc[:, 'Corpus'].to_list()
+
+
+# Setup the BM25 retriever
+def get_prompt(query, query_bm25):
+    prompt = 'به سوال کاربر بر اساس متن هایی که در ادامه آمده است پاسخ بدهید'  + '\n'
+    prompt += 'اگر قادر به جواب دادن به سوال نبودی، عبارت زیر را خروجی بده:' + '\n'
+    prompt += 'متاسفم در حال حاضر اطلاعات زیادی درباره سوال شما نمی دانم!' + '\n\n'
+    prompt += f'سوال:' + '\n' + query + '\n\n'
+    for idx, topic in enumerate(query_bm25):
+        prompt += f'متن {idx+1}: ' + topic + '\n'
+    
+    return prompt
+
+
+def get_relevant_topics(query, corpus, n=3):
+    # Build the best match 25 base
+    tokenized_corpus = [doc.split(" ") for doc in corpus]
+    bm25 = BM25Plus(tokenized_corpus)
+
+    tokenized_query = query.split(" ")
+    query_bm = bm25.get_top_n(tokenized_query, corpus, n=n)
+
+    return query_bm
+
+# Chatbot formatting
+if "messages" not in st.session_state:
+    st.session_state["messages"] = [{"role": "assistant", "content": "چطور می‌توانم به شما کمک کنم؟"}]
+
+if "drugs" not in st.session_state:
+    st.session_state["drugs"] = {"drug_name": [], "drug_link": []}
+
+for msg in st.session_state.messages:
+    st.chat_message(msg["role"]).write(msg["content"])
+
+
+if prompt := st.chat_input():
+    st.session_state.messages.append({"role": "user", "content": prompt})
+    st.chat_message("user").write(prompt)
+    topics = get_relevant_topics(prompt, corpus, n=5)
+    relevant_drugs = df.iloc[list(topics.values())]
+    st.sidebar.header('داروها')
+    for drug in relevant_drugs.iterrows():
+        drug_name = drug[1]['Farsi_generic_name']
+        drug_link = drug[1]['Link']
+        if drug_name not in st.session_state.drugs['drug_name']:
+            st.session_state.drugs['drug_name'].append(drug_name)
+            st.session_state.drugs['drug_link'].append(drug_link)
+
+    for i in range(len(st.session_state.drugs['drug_name'])):
+        if st.session_state.drugs['drug_name'][i] in prompt:
+            st.sidebar.link_button(st.session_state.drugs['drug_name'][i], st.session_state.drugs['drug_link'][i])
+
+    prompt = get_prompt(prompt, topics)
+    response = client.chat.completions.create(
+        #
+        # Required parameters
+        #
+        messages=[
+            # Set an optional system message. This sets the behavior of the
+            # assistant and can be used to provide specific instructions for
+            # how it should behave throughout the conversation.
+            {
+                "role": "system",
+                "content": "تو یک دستیار سودمند هستی."
+            },
+            # Set a user message for the assistant to respond to.
+            {
+                "role": "user",
+                "content": prompt,
+            }
+        ],
+
+        # The language model which will generate the completion.
+        model="llama3-70b-8192",
+        #
+        # Optional parameters
+    )
+    msg = response.choices[0].message.content
+    st.session_state.messages.append({"role": "assistant", "content": msg})
+    st.chat_message("assistant").write(msg)