Update app.py

app.py

@@ -118,12 +118,12 @@ def check_for_fragment(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)
+    match = re.match(r'<a href="([^"]+)#:~:text=([^"]+)" data-hash="([^"]+)">([^<]+) \((\d{4})\)</a>', citation_html)
     if match:
-        url, encoded_fragment, author, year = match.groups()
+        url, encoded_fragment, scc_hash, author, year = match.groups()
         fragment_text = urllib.parse.unquote(encoded_fragment)
-        return author, year, url, fragment_text
-    return None, None, None, None
+        return author, year, url, fragment_text, scc_hash
+    return None, None, None, None, None
 
 def parse_metadata_hash(metadata_html):
     # Parse metadata HTML to extract hash, username, task_name, date, time
@@ -136,6 +136,36 @@ def parse_metadata_hash(metadata_html):
             return scc_hash, username, task_name, date, time
     return None, None, None, None, None
 
+# --- JavaScript for extracting link attributes ---
+def verification_js():
+    return """
+    <script>
+    function extractCitationData() {
+        const citationInput = document.getElementById('citation_input');
+        const hashInput = document.getElementById('hash_input');
+        const citationOutput = document.getElementById('citation_output');
+        const hashOutput = document.getElementById('hash_output');
+
+        if (citationInput && hashInput && citationOutput && hashOutput) {
+            const citationLink = citationInput.getElementsByTagName('a')[0];
+            const hashLink = hashInput.getElementsByTagName('a')[0];
+
+            if (citationLink) {
+                citationOutput.value = citationLink.outerHTML;
+            } else {
+                citationOutput.value = '';
+            }
+
+            if (hashLink) {
+                hashOutput.value = hashLink.outerHTML;
+            } else {
+                hashOutput.value = '';
+            }
+        }
+    }
+    </script>
+    """
+
 # --- Live Clock JavaScript ---
 def live_clock():
     return """
@@ -308,27 +338,42 @@ with tabs[1]:
     st.header("Verify Citation")
     st.markdown("""
     <div class="info-card">
-    Paste the generated citation and hash with their embedded links below to verify the citation's authenticity.
+    Paste the generated citation and hash with their embedded links below to verify the citation's authenticity. Copy the rendered links directly from the output.
     </div>
     """, unsafe_allow_html=True)
 
     # Input fields for citation and hash
-    citation_input = st.text_area("Paste Citation (with embedded link)", help="Paste the citation HTML, e.g., <a href='URL'>Author (Year)</a>", placeholder='<a href="https://example.com#:~:text=fragment" data-hash="hash">Author (Year)</a>')
-    hash_input = st.text_area("Paste Hash (with embedded link)", help="Paste the hash HTML, e.g., <a href='URL'>Author (Year). Hash</a>", placeholder='<a href="https://example.com#:~:text=metadata" data-hash="hash">Author (Year). Hash</a>')
+    st.markdown("""
+    <div>
+        <label>Paste Citation (with embedded link)</label>
+        <div id="citation_input" contenteditable="true" style="border: 1px solid #e0e0e0; border-radius: 4px; padding: 0.5rem; min-height: 50px; margin-bottom: 1rem;"></div>
+        <input type="hidden" id="citation_output" name="citation_output">
+        <label>Paste Hash (with embedded link)</label>
+        <div id="hash_input" contenteditable="true" style="border: 1px solid #e0e0e0; border-radius: 4px; padding: 0.5rem; min-height: 50px; margin-bottom: 1rem;"></div>
+        <input type="hidden" id="hash_output" name="hash_output">
+    </div>
+    """, unsafe_allow_html=True)
+
+    # Add JavaScript to capture pasted link data
+    components.html(verification_js(), height=0)
+
+    # Capture the extracted data
+    citation_html = st.text_input("Citation HTML (auto-filled)", key="citation_output", label_visibility="collapsed")
+    hash_html = st.text_input("Hash HTML (auto-filled)", key="hash_output", label_visibility="collapsed")
 
     verify_button = st.button("Verify Citation", type="primary", use_container_width=True)
 
     if verify_button:
-        if not (citation_input and hash_input):
-            st.error("Please paste both the citation and hash before verifying.")
+        if not (citation_html and hash_html):
+            st.error("Please paste both the citation and hash links before verifying.")
         else:
             # Parse citation
-            author, year, url, fragment_text = parse_citation(citation_input)
+            author, year, url, fragment_text, citation_hash = parse_citation(citation_html)
             # Parse hash and metadata
-            scc_hash, username, task_name, date, time = parse_metadata_hash(hash_input)
+            scc_hash, username, task_name, date, time = parse_metadata_hash(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 formatted HTML with embedded links.")
+                st.error("Invalid citation or hash format. Please ensure both inputs are correctly pasted links from the generated output.")
             else:
                 # Recompute hash
                 recomputed_hash = generate_citation_hash(