Update app.py

app.py

@@ -137,9 +137,10 @@ def encode_text_fragment(text):
 def generate_citation_hash(author, year, url, fragment_text, cited_text, username, task_name, current_date, current_time):
     # Normalise hyphens for consistent hash generation
     normalised_fragment_text = normalise_hyphens(fragment_text)
-    normalised_cited_text = normalise_hyphens(cited_text)
-    normalised_task_name = normalise_hyphens(task_name)
-    data = f"{author}, {year} | {url} | {normalised_fragment_text} | {normalised_cited_text} | {username} | {normalised_task_name} | {current_date} | {current_time}"
+    # Use the longest segment for the cited text to align with citation link
+    selected_cited_text = select_longest_segment(cited_text)
+    normalised_cited_text = normalise_hyphens(selected_cited_text)
+    data = f"{author}, {year} | {url} | {normalised_fragment_text} | {normalised_cited_text} | {username} | {normalise_hyphens(task_name)} | {current_date} | {current_time}"
     return hashlib.sha256(data.encode('utf-8')).hexdigest()
 
 def format_citation_html(url, fragment_text, author, year, scc_hash):
@@ -451,20 +452,21 @@ with tabs[1]:
             hash_base_url, hash_fragment = parse_url(hash_url)
             # Parse metadata from hash URL fragment
             username, task_name, date, time = parse_metadata(hash_fragment)
-
+    
             if not all([author, year, citation_base_url, citation_fragment, scc_hash, hash_base_url, username, task_name, date, time]):
                 st.error("Invalid input format. Ensure the citation text, URLs, and SCC index text are correctly pasted from the generated output.")
             elif citation_base_url != hash_base_url:
                 st.error("The citation URL and SCC index URL must point to the same base URL.")
             else:
-                # Normalise hyphens for hash recomputation
+                # Normalise hyphens and select longest segment for hash recomputation
                 normalised_citation_fragment = normalise_hyphens(citation_fragment)
+                selected_citation_fragment = select_longest_segment(normalised_citation_fragment)
                 normalised_task_name = normalise_hyphens(task_name)
                 # Recompute hash
                 recomputed_hash = generate_citation_hash(
-                    author, year, citation_base_url, normalised_citation_fragment, normalised_citation_fragment, username, normalised_task_name, date, time
+                    author, year, citation_base_url, normalised_citation_fragment, selected_citation_fragment, username, normalised_task_name, date, time
                 )
-
+    
                 if recomputed_hash == scc_hash:
                     st.markdown("""
                     <div class="success-box">