Create app.py

app.py

@@ -0,0 +1,67 @@
+import streamlit as st
+from htmlrag import clean_html, build_block_tree
+from htmlrag import EmbedHTMLPruner
+from htmlrag import GenHTMLPruner
+import torch
+
+# Title
+st.title("HtmlRAG Demo - HTML Cleaning and Query Rewriting")
+
+# HTML document comparison before and after cleaning
+st.header("HTML Document Comparison Before and After Cleaning")
+
+# Upload an HTML document
+html_file = st.file_uploader("Upload an HTML file", type=["html"])
+
+if html_file is not None:
+    # Display the original HTML content
+    raw_html = html_file.getvalue().decode("utf-8")
+    st.subheader("Original HTML Content")
+    st.code(raw_html, language="html")
+    
+    # HtmlRAG cleaning
+    if st.button("Clean HTML"):
+        # Clean the HTML using the clean_html function
+        simplified_html = clean_html(raw_html)
+        st.subheader("Cleaned HTML Content")
+        st.code(simplified_html, language="html")
+        
+        # Compare the original HTML and the cleaned HTML
+        st.subheader("Comparison")
+        st.write("The parts removed or compressed will be highlighted in the cleaned version.")
+        
+        # Build the HTML block tree and display it
+        block_tree, simplified_html = build_block_tree(simplified_html, max_node_words=10)
+        st.subheader("Block Tree")
+        for block in block_tree:
+            st.write(f"Block Content: {block[0]}")
+            st.write(f"Block Path: {block[1]}")
+            st.write(f"Is Leaf: {block[2]}")
+            st.write("---")
+
+# Query rewriting and web search results visualization
+st.header("Query Rewriting and Web Search Results Visualization")
+
+# Input a query
+query = st.text_input("Enter a query:")
+
+if query:
+    # Display the rewritten query
+    rewritten_query = query
+    st.subheader("Rewritten Query")
+    st.write(rewritten_query)
+
+    # Simulate fetching search results
+    search_results = ["Result 1: Bellagio is a luxury hotel.", "Result 2: It was built in 1998.", "Result 3: The Bellagio is on the Las Vegas Strip."]
+    
+    # Display the search results
+    st.subheader("Search Results")
+    for i, result in enumerate(search_results):
+        st.write(f"Result {i+1}:")
+        st.write(result)
+        st.write("---")
+        
+    # Display a comparison of the original query and rewritten query
+    st.subheader("Query Comparison")
+    st.write(f"Original Query: {query}")
+    st.write(f"Rewritten Query: {rewritten_query}")