Update app.py

app.py

@@ -1,5 +1,4 @@
 import streamlit as st
-import streamlit.components.v1 as components
 import hashlib
 import urllib.parse
 from datetime import datetime
@@ -109,48 +108,46 @@ def format_metadata_html(url, author, year, scc_hash, username, task_name, curre
 def check_for_fragment(url):
     return '#:~:text=' in url
 
-def parse_citation(citation_html):
-    match = re.match(r'<a href="([^"]+)#:~:text=([^"]+)" data-hash="([^"]+)">([^<]+) \((\d{4})\)</a>', citation_html)
+def parse_citation_text(citation_text):
+    # Match "Author (Year)" or "(Author, Year)"
+    match = re.match(r'(\(?.*?\)?)\s*\((\d{4})\)', citation_text.strip())
     if match:
-        url, encoded_fragment, scc_hash, author, year = match.groups()
-        fragment_text = urllib.parse.unquote(encoded_fragment)
-        return author, year, url, fragment_text, scc_hash
-    return None, None, None, None, None
-
-def parse_metadata_hash(metadata_html):
-    match = re.match(r'<a href="([^"]+)#:~:text=([^"]+)" data-hash="([^"]+)">[^<]+\(\d{4}\)\. ([^<]+)</a>', metadata_html)
+        author, year = match.groups()
+        author = author.strip('()').strip()
+        return author, year
+    return None, None
+
+def parse_url(url):
+    if not url:
+        return None, None
+    try:
+        # Extract fragment text from URL
+        match = re.search(r'#:~:text=([^&]+)', url)
+        fragment_text = urllib.parse.unquote(match.group(1)) if match else None
+        # Extract base URL (before #)
+        base_url = url.split('#')[0]
+        return base_url, fragment_text
+    except:
+        return None, None
+
+def parse_hash_text(hash_text):
+    # Match "Author (Year). <hash>"
+    match = re.match(r'.*?\(\d{4}\)\.\s*([0-9a-f]{64})', hash_text.strip())
     if match:
-        url, encoded_metadata, scc_hash, same_hash = match.groups()
-        metadata_parts = urllib.parse.unquote(encoded_metadata).split('—')
+        return match.group(1)
+    return None
+
+def parse_metadata(fragment_text):
+    if not fragment_text:
+        return None, None, None, None
+    try:
+        metadata_parts = fragment_text.split('—')
         if len(metadata_parts) == 4:
             username, task_name, date, time = metadata_parts
-            return scc_hash, username, task_name, date, time
-    return None, None, None, None, None
-
-# --- JavaScript for capturing pasted links ---
-def verification_js():
-    return """
-    <script>
-    function updateCitationData(inputId, outputId) {
-        const input = document.getElementById(inputId);
-        const output = document.getElementById(outputId);
-        if (input && output) {
-            const link = input.querySelector('a');
-            output.value = link ? link.outerHTML : '';
-            const event = new Event('input', { bubbles: true });
-            output.dispatchEvent(event);
-        }
-    }
-
-    document.getElementById('citation_input').addEventListener('paste', function(e) {
-        setTimeout(() => updateCitationData('citation_input', 'citation_output'), 100);
-    });
-
-    document.getElementById('hash_input').addEventListener('paste', function(e) {
-        setTimeout(() => updateCitationData('hash_input', 'hash_output'), 100);
-    });
-    </script>
-    """
+            return username, task_name, date, time
+        return None, None, None, None
+    except:
+        return None, None, None, None
 
 # --- Live Clock JavaScript ---
 def live_clock():
@@ -176,7 +173,7 @@ def live_clock():
             const date = `${parts[4].value}-${parts[2].value}-${parts[0].value}`;
             const time = `${parts[6].value}:${parts[8].value}:${parts[10].value}`;
             const datetimeElement = document.getElementById('live_datetime');
-            if (datetimeElement) {
+            if datetimeElement) {
                 datetimeElement.innerText = `${date} ${time}`;
             }
         }
@@ -185,13 +182,6 @@ def live_clock():
     </script>
     """
 
-# --- Callbacks for updating session state ---
-def update_citation_html():
-    st.session_state.citation_html = st.session_state.citation_output
-
-def update_hash_html():
-    st.session_state.hash_html = st.session_state.hash_output
-
 # --- Streamlit App ---
 st.set_page_config(layout="wide", page_title="Smart Context Citation Tool")
 
@@ -274,7 +264,7 @@ with tabs[0]:
 
     # Live date and time display
     st.markdown("### Current Date and Time")
-    components.html(live_clock(), height=50)
+    st.components.v1.html(live_clock(), height=50)
 
     # Get current date and time in Melbourne timezone for hash generation
     melbourne_tz = pytz.timezone(MELBOURNE_TIMEZONE)
@@ -331,56 +321,44 @@ with tabs[1]:
     st.header("Verify Citation")
     st.markdown("""
     <div class="info-card">
-    Paste the rendered citation and hash links (e.g., "Author (Year)" and "Author (Year). Hash") directly from the 'Citation Generator' tab to verify the citation's authenticity.
+    Copy the citation text (e.g., "Abuseif et al. (2025)" or "(Abuseif et al., 2025)") and its hyperlink URL from the 'Citation Generator' tab. Similarly, copy the hash text (e.g., "Abuseif et al. (2025). <hash>") and its hyperlink URL. Paste them into the fields below to verify the citation's authenticity. To copy the URL, right-click the hyperlink and select "Copy Link Address".
     </div>
     """, unsafe_allow_html=True)
 
     # Input fields for citation and hash
-    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" value="">
-        <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" value="">
-    </div>
-    """, unsafe_allow_html=True)
-
-    # Add JavaScript to capture pasted link data
-    components.html(verification_js(), height=0)
+    st.subheader("Citation Information")
+    citation_text = st.text_input("Citation Text", help="Paste the citation text, e.g., 'Abuseif et al. (2025)' or '(Abuseif et al., 2025)'", placeholder="e.g., Abuseif et al. (2025)")
+    citation_url = st.text_input("Citation URL", help="Paste the hyperlink URL from the citation, e.g., 'https://example.com#:~:text=fragment'", placeholder="e.g., https://example.com#:~:text=fragment")
 
-    # Initialize session state
-    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 = ''
-    if 'citation_output' not in st.session_state:
-        st.session_state.citation_output = ''
-    if 'hash_output' not in st.session_state:
-        st.session_state.hash_output = ''
-
-    # Hidden inputs with on_change callbacks
-    st.text_input("Citation HTML", key="citation_output", value=st.session_state.citation_output, on_change=update_citation_html, label_visibility="collapsed")
-    st.text_input("Hash HTML", key="hash_output", value=st.session_state.hash_output, on_change=update_hash_html, label_visibility="collapsed")
+    st.subheader("Hash Information")
+    hash_text = st.text_input("Hash Text", help="Paste the hash text, e.g., 'Abuseif et al. (2025). <hash>'", placeholder="e.g., Abuseif et al. (2025). <hash>")
+    hash_url = st.text_input("Hash URL", help="Paste the hyperlink URL from the hash, e.g., 'https://example.com#:~:text=metadata'", placeholder="e.g., https://example.com#:~:text=metadata")
 
     verify_button = st.button("Verify Citation", type="primary", use_container_width=True)
 
     if verify_button:
-        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.")
+        if not all([citation_text, citation_url, hash_text, hash_url]):
+            st.error("Please provide all fields (citation text, citation URL, hash text, hash URL) before verifying.")
         else:
-            # Parse citation
-            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(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.")
+            # Parse citation text
+            author, year = parse_citation_text(citation_text)
+            # Parse citation URL
+            citation_base_url, citation_fragment = parse_url(citation_url)
+            # Parse hash text
+            scc_hash = parse_hash_text(hash_text)
+            # Parse hash URL
+            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 hash text are correctly pasted from the generated output.")
+            elif citation_base_url != hash_base_url:
+                st.error("The citation URL and hash URL must point to the same base URL.")
             else:
                 # Recompute hash
                 recomputed_hash = generate_citation_hash(
-                    author, year, url, fragment_text, fragment_text, username, task_name, date, time
+                    author, year, citation_base_url, citation_fragment, citation_fragment, username, task_name, date, time
                 )
 
                 if recomputed_hash == scc_hash: