Update app.py

app.py

@@ -4,7 +4,6 @@ import hashlib
 import urllib.parse
 from datetime import datetime
 import pytz
-import pandas as pd
 import re
 
 # --- Constants ---
@@ -84,12 +83,6 @@ def load_css():
         border-left: 1px solid #e0e0e0;
     }
     
-    .verification-table {
-        margin: 1rem 0;
-        border-radius: 4px;
-        overflow: hidden;
-    }
-    
     .rendered-citation {
         margin: 1rem 0;
         font-size: 1rem;
@@ -117,7 +110,6 @@ def check_for_fragment(url):
     return '#:~:text=' in url
 
 def parse_citation(citation_html):
-    # Parse citation HTML to extract author, year, URL, and fragment text
     match = re.match(r'<a href="([^"]+)#:~:text=([^"]+)" data-hash="([^"]+)">([^<]+) \((\d{4})\)</a>', citation_html)
     if match:
         url, encoded_fragment, scc_hash, author, year = match.groups()
@@ -126,7 +118,6 @@ def parse_citation(citation_html):
     return None, None, None, None, None
 
 def parse_metadata_hash(metadata_html):
-    # Parse metadata HTML to extract hash, username, task_name, date, time
     match = re.match(r'<a href="([^"]+)#:~:text=([^"]+)" data-hash="([^"]+)">[^<]+\(\d{4}\)\. ([^<]+)</a>', metadata_html)
     if match:
         url, encoded_metadata, scc_hash, same_hash = match.groups()
@@ -140,7 +131,7 @@ def parse_metadata_hash(metadata_html):
 def verification_js():
     return """
     <script>
-    function updateOutputs() {
+    function updateCitationData() {
         const citationInput = document.getElementById('citation_input');
         const hashInput = document.getElementById('hash_input');
         const citationOutput = document.getElementById('citation_output');
@@ -155,23 +146,12 @@ def verification_js():
         }
     }
 
-    document.addEventListener('DOMContentLoaded', function() {
-        const citationInput = document.getElementById('citation_input');
-        const hashInput = document.getElementById('hash_input');
-        
-        if (citationInput) {
-            citationInput.addEventListener('paste', function() {
-                setTimeout(updateOutputs, 100); // Delay to ensure paste is complete
-            });
-            citationInput.addEventListener('input', updateOutputs);
-        }
-        
-        if (hashInput) {
-            hashInput.addEventListener('paste', function() {
-                setTimeout(updateOutputs, 100); // Delay to ensure paste is complete
-            });
-            hashInput.addEventListener('input', updateOutputs);
-        }
+    document.getElementById('citation_input').addEventListener('paste', function(e) {
+        setTimeout(updateCitationData, 0);
+    });
+
+    document.getElementById('hash_input').addEventListener('paste', function(e) {
+        setTimeout(updateCitationData, 0);
     });
     </script>
     """
@@ -352,12 +332,6 @@ with tabs[1]:
     </div>
     """, unsafe_allow_html=True)
 
-    # Initialize session state for storing extracted data
-    if 'citation_html' not in st.session_state:
-        st.session_state.citation_html = ''
-    if 'hash_html' not in st.session_state:
-        st.session_state.hash_html = ''
-
     # Input fields for citation and hash
     st.markdown("""
     <div>
@@ -373,20 +347,49 @@ with tabs[1]:
     # Add JavaScript to capture pasted link data
     components.html(verification_js(), height=0)
 
-    # Use session state to capture JavaScript output
-    citation_html = st.session_state.get('citation_output', '')
-    hash_html = st.session_state.get('hash_output', '')
+    # Use session state to store the extracted data
+    if 'citation_html' not in st.session_state:
+        st.session_state.citation_html = ''
+    if 'hash_html' not in st.session_state:
+        st.session_state.hash_html = ''
+
+    # JavaScript to update session state via URL parameters
+    st.markdown("""
+    <script>
+        function updateSessionState() {
+            const citationOutput = document.getElementById('citation_output').value;
+            const hashOutput = document.getElementById('hash_output').value;
+            const params = new URLSearchParams(window.location.search);
+            params.set('citation_html', encodeURIComponent(citationOutput));
+            params.set('hash_html', encodeURIComponent(hashOutput));
+            window.history.replaceState({}, '', `${window.location.pathname}?${params}`);
+        }
+        document.getElementById('citation_input').addEventListener('paste', function() {
+            setTimeout(updateCitationData, 0);
+            setTimeout(updateSessionState, 100);
+        });
+        document.getElementById('hash_input').addEventListener('paste', function() {
+            setTimeout(updateCitationData, 0);
+            setTimeout(updateSessionState, 100);
+        });
+    </script>
+    """, unsafe_allow_html=True)
+
+    # Retrieve data from URL parameters
+    query_params = st.experimental_get_query_params()
+    st.session_state.citation_html = query_params.get('citation_html', [''])[0]
+    st.session_state.hash_html = query_params.get('hash_html', [''])[0]
 
     verify_button = st.button("Verify Citation", type="primary", use_container_width=True)
 
     if verify_button:
-        if not (citation_html and hash_html):
+        if not (st.session_state.citation_html and st.session_state.hash_html):
             st.error("Please paste both the citation and hash links before verifying.")
         else:
             # Parse citation
-            author, year, url, fragment_text, citation_hash = parse_citation(citation_html)
+            author, year, url, fragment_text, citation_hash = parse_citation(st.session_state.citation_html)
             # Parse hash and metadata
-            scc_hash, username, task_name, date, time = parse_metadata_hash(hash_html)
+            scc_hash, username, task_name, date, time = parse_metadata_hash(st.session_state.hash_html)
 
             if not all([author, year, url, fragment_text, scc_hash, username, task_name, date, time]):
                 st.error("Invalid citation or hash format. Please ensure both inputs are correctly pasted links from the generated output.")