""" Individual Record Update View Provides record-by-record field comparison and update controls. """ import streamlit as st import pandas as pd from typing import List, Dict, Any from services.update_engine import UpdateEngine from bson import ObjectId def _render_individual_record_updates( duplicates: List[Dict[str, Any]], collection_name: str, mongo_service, update_engine: UpdateEngine ): """ Render individual record-level field comparison and updates. Args: duplicates: List of duplicate records collection_name: Name of the collection mongo_service: MongoService instance update_engine: UpdateEngine instance """ st.markdown("### 📝 Individual Record Updates") st.info(f""" **Individual Record Mode** - Review each record separately - Select fields to update for each record - Apply updates per record or in bulk Total records: **{len(duplicates)}** """) # Initialize session state for individual updates if 'individual_update_decisions' not in st.session_state: st.session_state.individual_update_decisions = {} # Record selector st.markdown("#### Select Record to Review") col1, col2, col3 = st.columns([2, 1, 1]) with col1: # Create record options with business key info record_options = [] for idx, dup in enumerate(duplicates): match_method = dup.get('match_method', 'unknown') matching_key = dup.get('matching_key', 'N/A') record_options.append(f"Record #{idx + 1}: {matching_key}") selected_record_idx = st.selectbox( "Choose a record", options=range(len(duplicates)), format_func=lambda i: record_options[i], key='individual_record_selector' ) with col2: st.metric("Current Record", f"{selected_record_idx + 1}/{len(duplicates)}") with col3: # Navigation buttons col_prev, col_next = st.columns(2) with col_prev: if st.button("⬅️ Prev", disabled=selected_record_idx == 0, use_container_width=True): st.session_state.individual_record_selector = selected_record_idx - 1 st.rerun() with col_next: if st.button("Next ➡️", disabled=selected_record_idx == len(duplicates) - 1, use_container_width=True): st.session_state.individual_record_selector = selected_record_idx + 1 st.rerun() st.markdown("---") # Get selected record selected_dup = duplicates[selected_record_idx] json_record = selected_dup['json_record'] db_record = selected_dup['db_record'] # Show match info match_method = selected_dup.get('match_method', 'unknown') matching_key = selected_dup.get('matching_key', 'N/A') match_icon = "🔑" if match_method == 'primary_key' else "🔍" match_label = "Primary Key" if match_method == 'primary_key' else "Alternative Key" st.info(f"{match_icon} **Matched by {match_label}:** {matching_key}") # Compare fields from services.import_preview_engine import ImportPreviewEngine preview_engine = ImportPreviewEngine() # Build comparison for this single record comp_df = preview_engine.build_duplicate_comparison_dataframe([selected_dup]) if comp_df.empty: st.warning("No field differences found") return # Initialize decisions for this record record_key = f"record_{selected_record_idx}" if record_key not in st.session_state.individual_update_decisions: # Default: update new and different fields default_decisions = {} for _, row in comp_df.iterrows(): field = row['field'] status = row['status'] if status in ['new', 'different']: default_decisions[field] = 'update' else: default_decisions[field] = 'skip' st.session_state.individual_update_decisions[record_key] = default_decisions # Quick action buttons for this record st.markdown("#### Quick Actions for This Record") col1, col2, col3, col4 = st.columns(4) with col1: if st.button("✅ Update All", key=f"update_all_{selected_record_idx}", use_container_width=True): for field in st.session_state.individual_update_decisions[record_key].keys(): st.session_state.individual_update_decisions[record_key][field] = 'update' st.rerun() with col2: if st.button("❌ Skip All", key=f"skip_all_{selected_record_idx}", use_container_width=True): for field in st.session_state.individual_update_decisions[record_key].keys(): st.session_state.individual_update_decisions[record_key][field] = 'skip' st.rerun() with col3: if st.button("✅ Update New", key=f"update_new_{selected_record_idx}", use_container_width=True): for _, row in comp_df.iterrows(): field = row['field'] if row['status'] == 'new': st.session_state.individual_update_decisions[record_key][field] = 'update' st.rerun() with col4: if st.button("✅ Update Different", key=f"update_diff_{selected_record_idx}", use_container_width=True): for _, row in comp_df.iterrows(): field = row['field'] if row['status'] == 'different': st.session_state.individual_update_decisions[record_key][field] = 'update' st.rerun() st.markdown("---") # Display field comparison with checkboxes st.markdown("#### Field-by-Field Comparison") # Add decision column to dataframe comp_df['_update_decision'] = comp_df['field'].apply( lambda f: st.session_state.individual_update_decisions[record_key].get(f, 'skip') ) # Color coding function def highlight_row(row): styles = [] status = row.get('status', '') for col in comp_df.columns: if col == 'field': styles.append('font-weight: bold; background-color: #f5f5f5') elif col == 'status': if status == 'new': styles.append('background-color: #4CAF50; color: white; font-weight: bold') elif status == 'different': styles.append('background-color: #FF9800; color: white; font-weight: bold') elif status == 'missing_in_json': styles.append('background-color: #f44336; color: white; font-weight: bold') elif status == 'match': styles.append('background-color: #9E9E9E; color: white') else: styles.append('') elif col == 'json_value': if status == 'new': styles.append('background-color: #C8E6C9; color: #1B5E20; font-weight: bold') elif status == 'different': styles.append('background-color: #BBDEFB; color: #0D47A1; font-weight: bold') else: styles.append('background-color: #E3F2FD; color: #1565C0') elif col == 'db_value': if status == 'different': styles.append('background-color: #FFE0B2; color: #E65100; font-weight: bold') elif status == 'missing_in_json': styles.append('background-color: #FFCDD2; color: #B71C1C; font-weight: bold') else: styles.append('background-color: #FFF3E0; color: #E65100') else: styles.append('') return styles # Display styled dataframe styled_df = comp_df.style.apply(highlight_row, axis=1) st.dataframe( styled_df, use_container_width=True, hide_index=True, height=400 ) st.markdown(""" **🎨 Color Legend:** - 🟢 **NEW** (Green) - Field exists in JSON but NOT in MongoDB - 🟠 **DIFFERENT** (Orange) - Field value differs - 🔴 **MISSING** (Red) - Field exists in MongoDB but NOT in JSON - ⚪ **MATCH** (Gray) - Values are identical """) # Editable field selection st.markdown("#### Select Fields to Update") # Create checkboxes for each field update_count = 0 for _, row in comp_df.iterrows(): field = row['field'] status = row['status'] json_val = row['json_value'] db_val = row['db_value'] current_decision = st.session_state.individual_update_decisions[record_key].get(field, 'skip') col1, col2, col3, col4 = st.columns([3, 2, 2, 1]) with col1: st.text(f"📌 {field}") with col2: st.text(f"JSON: {json_val}") with col3: st.text(f"DB: {db_val}") with col4: # Status badge if status == 'new': badge = "🟢 NEW" elif status == 'different': badge = "🟠 DIFF" elif status == 'match': badge = "⚪ MATCH" else: badge = "🔴 MISS" update_field = st.checkbox( badge, value=current_decision == 'update', key=f"field_update_{selected_record_idx}_{field}", label_visibility="visible" ) if update_field: st.session_state.individual_update_decisions[record_key][field] = 'update' update_count += 1 else: st.session_state.individual_update_decisions[record_key][field] = 'skip' st.markdown("---") # Summary and apply button st.markdown("#### Update Summary") col1, col2, col3 = st.columns(3) with col1: st.metric("Fields to Update", update_count) with col2: st.metric("Total Fields", len(comp_df)) with col3: # Count total updates across all records total_updates = sum( sum(1 for decision in decisions.values() if decision == 'update') for decisions in st.session_state.individual_update_decisions.values() ) st.metric("Total Updates (All Records)", total_updates) # Apply updates button if update_count > 0: st.warning(f"⚠️ Ready to update {update_count} field(s) in this record") col1, col2 = st.columns([1, 3]) with col1: if st.button("🚀 Apply Updates to This Record", type="primary", use_container_width=True): _apply_individual_record_update( selected_dup, st.session_state.individual_update_decisions[record_key], collection_name, mongo_service ) with col2: if st.button("🚀 Apply All Updates (All Records)", type="secondary", use_container_width=True): _apply_all_individual_updates( duplicates, st.session_state.individual_update_decisions, collection_name, mongo_service ) def _apply_individual_record_update( duplicate_record: Dict[str, Any], field_decisions: Dict[str, str], collection_name: str, mongo_service ): """Apply updates to a single record.""" try: # Get fields to update fields_to_update = { field: duplicate_record['json_record'][field] for field, decision in field_decisions.items() if decision == 'update' and field in duplicate_record['json_record'] } if not fields_to_update: st.warning("No fields selected for update") return # Get collection collection = mongo_service.get_collection(collection_name) if collection is None: st.error("❌ Collection not found") return # Prepare update db_id = duplicate_record['db_record']['_id'] with st.spinner("Updating record..."): result = collection.update_one( {'_id': ObjectId(str(db_id))}, {'$set': fields_to_update} ) if result.modified_count > 0: st.success(f"✅ Successfully updated {len(fields_to_update)} field(s)") # Clear this record's decisions st.session_state.individual_update_decisions = {} st.rerun() else: st.warning("⚠️ No changes were made") except Exception as e: st.error(f"❌ Update failed: {str(e)}") def _apply_all_individual_updates( duplicates: List[Dict[str, Any]], all_decisions: Dict[str, Dict[str, str]], collection_name: str, mongo_service ): """Apply updates to all records with pending decisions.""" try: collection = mongo_service.get_collection(collection_name) if collection is None: st.error("❌ Collection not found") return updated_count = 0 errors = [] with st.spinner("Updating all records..."): for idx, duplicate_record in enumerate(duplicates): record_key = f"record_{idx}" if record_key not in all_decisions: continue field_decisions = all_decisions[record_key] # Get fields to update fields_to_update = { field: duplicate_record['json_record'][field] for field, decision in field_decisions.items() if decision == 'update' and field in duplicate_record['json_record'] } if not fields_to_update: continue try: db_id = duplicate_record['db_record']['_id'] result = collection.update_one( {'_id': ObjectId(str(db_id))}, {'$set': fields_to_update} ) if result.modified_count > 0: updated_count += 1 except Exception as e: errors.append(f"Record #{idx + 1}: {str(e)}") if updated_count > 0: st.success(f"✅ Successfully updated {updated_count} record(s)") if errors: st.warning(f"⚠️ {len(errors)} update(s) failed:") for error in errors[:5]: st.text(f" - {error}") # Clear all decisions st.session_state.individual_update_decisions = {} st.session_state.comparison_results = None st.session_state.comparison_loaded = False st.info("💡 Reload the comparison to see updated results") st.rerun() else: st.warning("⚠️ No records were updated") if errors: for error in errors: st.error(error) except Exception as e: st.error(f"❌ Failed to update: {str(e)}")