Update app.py

app.py

@@ -32,7 +32,9 @@ st.info("""This Application currently only works with arxiv and acl anthology we
 1) Arxiv:- https://arxiv.org/abs/paper_unique_identifier
 2) ACL Anthology:- https://aclanthology.org/paper_unique_identifier/ 
 
-This Application uses the recently released Meta Nougat Visual Transformer for processing Papers""", icon="ℹ️")
+This Application uses the recently released Meta Nougat Visual Transformer for processing Papers.
+ 
+The Nougat Transformer is inferenced through a deployed app I created on the Modal platform(https://modal.com/) and uses T4 GPU as hardware""", icon="ℹ️")
 user_input = st.text_input("Enter the arxiv or acl anthology url of the paper", "https://aclanthology.org/2023.semeval-1.266/")
 
 
@@ -55,16 +57,33 @@ def initialize_session_state():
 initialize_session_state()
 
 
+def is_arxiv_url(url: str) -> bool:
+    import re
+
+    arxiv_pattern = r'https?://arxiv\.org/abs/.+'
+    return bool(re.match(arxiv_pattern, url))
+
+
+def is_acl_anthology_url(url: str) -> bool:
+    import re
+
+    acl_anthology_pattern = r'https://aclanthology\.org/.*?/'
+    return bool(re.match(acl_anthology_pattern, url))
+
+
 def get_paper_content(url: str) -> str:
-    with st.spinner(text="Using Nougat(https://facebookresearch.github.io/nougat/) to read the paper contents and get the markdown representation of the paper"):
-        f = modal.Function.lookup("streamlit-hack", "main")
-        output = f.call(url)
-        st.session_state.paper_content = output
-        return output
+    with st.spinner(text="Using Nougat(https://facebookresearch.github.io/nougat/) to read the paper contents and get the markdown representation of the paper  – hang tight! This should take 1-2 minutes"):
+        if is_arxiv_url(url=url) or is_acl_anthology_url(url=url):
+            f = modal.Function.lookup("streamlit-hack", "main")
+            output = f.call(url)
+            st.session_state.paper_content = output
+            return output
+        else:
+            return 'Invalid URL. Please provide a valid ArXiv or ACL Anthology URL.'
 
 
 def index_paper_content(content: str):
-    with st.spinner(text="Indexing the paper – hang tight! This should take 3-5 minutes"):
+    with st.spinner(text="Indexing the paper – hang tight! This should take 1-2 minutes"):
         try:
             LLM_USER_ID = 'openai'
             LLM_APP_ID = 'chat-completion'
@@ -95,7 +114,7 @@ def index_paper_content(content: str):
 
 
 def generate_insights():
-    with st.spinner(text="Generating insights on the paper and preparing the Chatbot"):
+    with st.spinner(text="Generating insights on the paper and preparing the Chatbot. Hang tight! this should take 1-2 mins."):
         try:
             LLM_USER_ID = 'openai'
             LLM_APP_ID = 'chat-completion'
@@ -133,28 +152,28 @@ def generate_insights():
             )
 
             response_key_insights = query_engine.query("Generate core crux insights, contributions and results of the paper as Key Topics and thier content in markdown format where each Key Topic is in bold followed by its content")
-
+            st.session_state.paper_insights = response_key_insights.response
         except Exception as e:
             print(str(e))
             response_key_insights = "Error While Generating Insights"
-
-        st.session_state.paper_insights = response_key_insights.response
+            st.session_state.paper_insights = response_key_insights
 
 
 if st.button("Read and Index Paper"):
     paper_content = get_paper_content(url=user_input)
 
-    if st.session_state.paper_content is not None:
-        with st.expander("See Paper Contents"):
-            st.markdown(paper_content)
+    if 'Invalid URL. Please provide a valid ArXiv or ACL Anthology URL.' in paper_content:
+        st.write('Invalid URL. Please provide a valid ArXiv or ACL Anthology URL.')
 
-        result = index_paper_content(content=paper_content)
-        st.write(result)
-        generate_insights()
+    else:
+        if st.session_state.paper_content is not None:
+            result = index_paper_content(content=paper_content)
+            st.write(result)
+            generate_insights()
 
 
 if st.session_state.paper_content is not None:
-    with st.expander("See Paper Contents"):
+    with st.expander("See Research Paper Contents"):
         st.markdown(st.session_state.paper_content)
 
 if st.session_state.paper_insights is not None:
@@ -292,5 +311,5 @@ if st.session_state.vector_store is not None:
                 else:
                     response = f"This query cannot be processed as it has been detected to be {reason}"
                     st.write(response)
-                    message = {"role": "assistant", "content": response.response}
+                    message = {"role": "assistant", "content": response}
                     st.session_state.messages.append(message)