revert to only pdfs

app.py

@@ -32,21 +32,9 @@ def pdf_to_document_store(pdf_file):
         preprocessed_docs=preprocessor.process(doc)
         document_store.write_documents(preprocessed_docs)
         temp_file.close()
-
-def crawl_url(url):
-    crawler = Crawler(output_dir="crawled_files", overwrite_existing_files=True, crawler_depth=1)
-    try:
-        docs = crawler.crawl(urls=[url])
-        preprocessed_docs = preprocessor.process(docs)
-        document_store.write_documents(preprocessed_docs)
-    except:
-        st.write('We were unable to crawl the contents of that URL, please try something else')
     
 def summarize(content):
-    if st.session_state.pdf:
-        pdf_to_document_store(content)
-    elif st.session_state.url:
-        crawl_url(content)
+    pdf_to_document_store(content)
     summaries = summarizer.predict(documents=document_store.get_all_documents(), generate_single_summary=True)
     return summaries
 
@@ -55,8 +43,6 @@ def set_state_if_absent(key, value):
         st.session_state[key] = value
         
 set_state_if_absent("summaries", None)
-set_state_if_absent("url", False)
-set_state_if_absent("pdf", False)
         
 document_store, summarizer, preprocessor = start_haystack()
 
@@ -69,24 +55,11 @@ This Summarization demo uses a [Haystack TransformerSummarizer node](https://hay
 """, unsafe_allow_html=True)
 
 uploaded_file = st.file_uploader("Choose a PDF file", accept_multiple_files=False)
-url = st.text_input(label="enter a URL")
-
-if (validators.url(url)) and (uploaded_file is None):
-    if st.button('Summarize contents of URL'):
-        with st.spinner("📚    Please wait while we produce a summary..."):
-            try:
-                st.session_state.pdf = False
-                st.session_state.url = True
-                st. session_state.summaries = summarize(url)
-            except Exception as e:
-                logging.exception(e)
                 
-if (uploaded_file is not None) and not validators.url(url):
+if uploaded_file is not None :
     if st.button('Summarize Document'):
         with st.spinner("📚    Please wait while we produce a summary..."):
             try:
-                st.session_state.pdf = True
-                st.session_state.url = False
                 st.session_state.summaries = summarize(uploaded_file)
             except Exception as e:
                 logging.exception(e)