Update 2026-01-29 00:53:13

app.py

@@ -88,6 +88,26 @@ def parse_bibtex_input(bibtex_content):
 
     return pd.DataFrame(papers).drop_duplicates(subset="Key", keep="first").reset_index(drop=True)
 
+def clean_bibtex_fields(bibtex):
+    """Remove unwanted fields from BibTeX entries"""
+    # Fields to remove
+    fields_to_remove = ['url', 'source', 'publication_stage', 'note']
+    
+    for field in fields_to_remove:
+        # Remove field with its value (handles multi-line values)
+        bibtex = re.sub(
+            rf'\s*{field}\s*=\s*\{{[^}}]*\}},?\s*',
+            '\n ',
+            bibtex,
+            flags=re.IGNORECASE
+        )
+    
+    # Clean up extra commas and whitespace
+    bibtex = re.sub(r',\s*,', ',', bibtex)
+    bibtex = re.sub(r',(\s*)\}', r'\1}', bibtex)
+    
+    return bibtex
+
 def protect_acronyms_in_fields(bibtex):
     """Protect acronyms with braces"""
     def clean_field(field_name, text):
@@ -204,6 +224,15 @@ def add_journal_abbreviations(df):
             row_data.get('Crossref_BibTeX_Abbrev', row['BibTeX'])
         )
         
+        # Clean unwanted fields from all BibTeX versions
+        row_data['BibTeX'] = clean_bibtex_fields(row_data['BibTeX'])
+        if row_data.get('Crossref_BibTeX'):
+            row_data['Crossref_BibTeX'] = clean_bibtex_fields(row_data['Crossref_BibTeX'])
+        if row_data.get('Crossref_BibTeX_Abbrev'):
+            row_data['Crossref_BibTeX_Abbrev'] = clean_bibtex_fields(row_data['Crossref_BibTeX_Abbrev'])
+        if row_data.get('Crossref_BibTeX_Protected'):
+            row_data['Crossref_BibTeX_Protected'] = clean_bibtex_fields(row_data['Crossref_BibTeX_Protected'])
+        
         abbreviated_rows.append(row_data)
 
     return pd.DataFrame(abbreviated_rows)
@@ -413,6 +442,23 @@ def export_database():
     except Exception as e:
         return jsonify({'error': str(e)}), 500
 
+@app.route('/api/database/download', methods=['GET'])
+def download_database():
+    """Download the SQLite database file"""
+    try:
+        db_path = app.config['DATABASE']
+        if not os.path.exists(db_path):
+            return jsonify({'error': 'Database file not found'}), 404
+        
+        return send_file(
+            db_path,
+            mimetype='application/x-sqlite3',
+            as_attachment=True,
+            download_name='refs_management.db'
+        )
+    except Exception as e:
+        return jsonify({'error': str(e)}), 500
+
 @app.route('/api/database/export-bibtex', methods=['GET'])
 def export_bibtex():
     """Export database as BibTeX"""

templates/index.html

@@ -514,6 +514,7 @@
                     <button class="export-btn" onclick="loadDatabaseEntries()">🔄 Refresh</button>
                     <button class="export-btn" onclick="exportCSV()">📥 Export CSV</button>
                     <button class="export-btn" onclick="exportBibTeX()">📥 Export BibTeX</button>
+                    <button class="export-btn" onclick="downloadDatabase()">💾 Download Database</button>
                 </div>
 
                 <div id="databaseEntries"></div>
@@ -660,7 +661,10 @@
             const bibtexPreview = document.getElementById('bibtexPreview');
             const jsonPreview = document.getElementById('jsonPreview');
 
-            resultsInfo.innerHTML = `<strong>✅ Processed ${data.count} references</strong>`;
+            const previewLimit = 10;
+            const hasMore = data.count > previewLimit;
+            
+            resultsInfo.innerHTML = `<strong>✅ Processed ${data.count} references</strong>${hasMore ? ` <span style="color: #666;">(Showing first ${previewLimit} in preview)</span>` : ''}`;
 
             // Table view
             let tableHtml = '<table class="results-table"><thead><tr>';
@@ -670,7 +674,8 @@
             });
             tableHtml += '</tr></thead><tbody>';
 
-            data.data.forEach(row => {
+            const displayData = data.data.slice(0, previewLimit);
+            displayData.forEach(row => {
                 tableHtml += '<tr>';
                 tableHtml += `<td><span class="badge badge-info">${row.Key}</span></td>`;
                 tableHtml += `<td>${row.Type}</td>`;
@@ -680,21 +685,36 @@
                 tableHtml += '</tr>';
             });
 
+            if (hasMore) {
+                tableHtml += `<tr><td colspan="5" style="text-align: center; padding: 20px; color: #666; font-style: italic;">... and ${data.count - previewLimit} more references (check Database tab or export to see all)</td></tr>`;
+            }
+
             tableHtml += '</tbody></table>';
             resultsTable.innerHTML = tableHtml;
 
-            // BibTeX view
+            // BibTeX view - limit to first few entries
             let bibtexText = '';
-            data.full_data.forEach(row => {
+            const displayFullData = data.full_data.slice(0, previewLimit);
+            displayFullData.forEach(row => {
                 const bibField = document.getElementById('protectCheckbox').checked ? 
                     'Crossref_BibTeX_Protected' : 
                     'Crossref_BibTeX_Abbrev';
                 bibtexText += (row[bibField] || row.BibTeX) + '\n\n';
             });
+            
+            if (hasMore) {
+                bibtexText += `\n... and ${data.count - previewLimit} more references\n(Export BibTeX from Database tab to get all entries)`;
+            }
+            
             bibtexPreview.textContent = bibtexText;
 
-            // JSON view
-            jsonPreview.textContent = JSON.stringify(data.full_data, null, 2);
+            // JSON view - limit preview
+            const jsonData = hasMore ? data.full_data.slice(0, previewLimit) : data.full_data;
+            let jsonText = JSON.stringify(jsonData, null, 2);
+            if (hasMore) {
+                jsonText += `\n\n// ... and ${data.count - previewLimit} more references`;
+            }
+            jsonPreview.textContent = jsonText;
         }
 
         async function loadDatabaseEntries() {
@@ -770,6 +790,10 @@
             window.location.href = '/api/database/export-bibtex';
         }
 
+        async function downloadDatabase() {
+            window.location.href = '/api/database/download';
+        }
+
         function clearInput() {
             document.getElementById('bibtexInput').value = '';
         }
@@ -784,4 +808,4 @@
         loadStats();
     </script>
 </body>
-</html>
+</html>