Update updated_word.py

updated_word.py

@@ -442,6 +442,147 @@ def semantic_text_matching(text, flat_json):
     
     return None
 
+def handle_nature_of_business_section(table, flat_json):
+    """TARGETED FIX for Issue 1: Nature of Business multi-line and sub-fields"""
+    replacements_made = 0
+    
+    for row_idx, row in enumerate(table.rows):
+        if len(row.cells) >= 1:
+            cell = row.cells[0]
+            cell_text = get_clean_text(cell).lower()
+            
+            # Check if this is the "Nature of the Operators Business" section
+            if "nature of the operators business" in cell_text or "nature of the operator business" in cell_text:
+                print(f"    🎯 Found Nature of Business section in row {row_idx + 1}")
+                
+                # Handle main business description (multi-line red text)
+                if has_red_text(cell):
+                    # Try to find business description in JSON
+                    business_desc_keys = [
+                        "nature of the operators business",
+                        "business description", 
+                        "operator business summary",
+                        "business summary"
+                    ]
+                    
+                    business_value = None
+                    for key in business_desc_keys:
+                        business_value = find_matching_json_value(key, flat_json)
+                        if business_value:
+                            break
+                    
+                    if business_value:
+                        business_text = get_value_as_string(business_value)
+                        cell_replacements = replace_red_text_in_cell(cell, business_text)
+                        replacements_made += cell_replacements
+                        print(f"      ✅ Updated main business description")
+                
+                # Look for sub-fields in the next few rows
+                for sub_row_idx in range(row_idx + 1, min(row_idx + 4, len(table.rows))):
+                    sub_row = table.rows[sub_row_idx]
+                    if len(sub_row.cells) >= 1:
+                        sub_cell = sub_row.cells[0]
+                        sub_text = get_clean_text(sub_cell).lower()
+                        
+                        # Handle Accreditation Number
+                        if "accreditation number" in sub_text and has_red_text(sub_cell):
+                            accred_value = find_matching_json_value("accreditation number", flat_json)
+                            if not accred_value:
+                                accred_value = find_matching_json_value("nhvas accreditation no", flat_json)
+                            if accred_value:
+                                accred_text = get_value_as_string(accred_value)
+                                cell_replacements = replace_red_text_in_cell(sub_cell, accred_text)
+                                replacements_made += cell_replacements
+                                print(f"      ✅ Updated Accreditation Number: {accred_text}")
+                        
+                        # Handle Expiry Date
+                        elif "expiry date" in sub_text and has_red_text(sub_cell):
+                            expiry_value = find_matching_json_value("expiry date", flat_json)
+                            if not expiry_value:
+                                expiry_value = find_matching_json_value("accreditation expiry", flat_json)
+                            if expiry_value:
+                                expiry_text = get_value_as_string(expiry_value)
+                                cell_replacements = replace_red_text_in_cell(sub_cell, expiry_text)
+                                replacements_made += cell_replacements
+                                print(f"      ✅ Updated Expiry Date: {expiry_text}")
+                
+                break  # Found the section, no need to continue
+    
+    return replacements_made
+
+
+def handle_operator_declaration_table(table, flat_json):
+    """TARGETED FIX for Issue 2: Operator Declaration Print Name and Position Title"""
+    replacements_made = 0
+    
+    for row_idx, row in enumerate(table.rows):
+        if len(row.cells) >= 2:
+            cell1_text = get_clean_text(row.cells[0]).lower()
+            cell2_text = get_clean_text(row.cells[1]).lower()
+            
+            # Check if this is the header row with "Print Name" and "Position Title"
+            if "print name" in cell1_text and ("position title" in cell2_text or "position" in cell2_text):
+                print(f"    🎯 Found Operator Declaration header row {row_idx + 1}")
+                
+                # Look for the data row (next row with red text)
+                if row_idx + 1 < len(table.rows):
+                    data_row = table.rows[row_idx + 1]
+                    if len(data_row.cells) >= 2:
+                        name_cell = data_row.cells[0]
+                        position_cell = data_row.cells[1]
+                        
+                        # Handle Print Name (first column)
+                        if has_red_text(name_cell):
+                            # Try to find operator name
+                            name_keys = [
+                                "operator name",
+                                "print name", 
+                                "legal entity",
+                                "operator"
+                            ]
+                            
+                            name_value = None
+                            for key in name_keys:
+                                name_value = find_matching_json_value(key, flat_json)
+                                if name_value:
+                                    break
+                            
+                            if name_value:
+                                name_text = get_value_as_string(name_value)
+                                cell_replacements = replace_red_text_in_cell(name_cell, name_text)
+                                replacements_made += cell_replacements
+                                print(f"      ✅ Updated Print Name: {name_text}")
+                        
+                        # Handle Position Title (second column)
+                        if has_red_text(position_cell):
+                            # Try to find position/title
+                            position_keys = [
+                                "position title",
+                                "position", 
+                                "title",
+                                "job title",
+                                "role"
+                            ]
+                            
+                            position_value = None
+                            for key in position_keys:
+                                position_value = find_matching_json_value(key, flat_json)
+                                if position_value:
+                                    break
+                            
+                            # If no specific position found, default to "Manager"
+                            if not position_value:
+                                position_value = "Manager"
+                            
+                            position_text = get_value_as_string(position_value)
+                            cell_replacements = replace_red_text_in_cell(position_cell, position_text)
+                            replacements_made += cell_replacements
+                            print(f"      ✅ Updated Position Title: {position_text}")
+                
+                break  # Found the section, no need to continue
+    
+    return replacements_made
+
 def handle_australian_company_number(row, company_numbers):
     """Enhanced ACN handling"""
     replacements_made = 0
@@ -688,6 +829,21 @@ def process_tables(document, flat_json):
             print_accreditation_replacements = handle_print_accreditation_section(table, flat_json)
             replacements_made += print_accreditation_replacements
             continue
+
+        # 🎯 NEW: TARGETED FIX 1 - Nature of Business section
+        if "nature of the operator" in table_text:
+            print(f"    🎯 Detected Nature of Business section")
+            nature_replacements = handle_nature_of_business_section(table, flat_json)
+            replacements_made += nature_replacements
+            # Don't continue - let it fall through to regular processing too
+        
+        # 🎯 NEW: TARGETED FIX 2 - Operator Declaration table  
+        if "print name" in table_text and "position title" in table_text:
+            print(f"    🎯 Detected Operator Declaration table")
+            declaration_replacements = handle_operator_declaration_table(table, flat_json)
+            replacements_made += declaration_replacements
+            # Don't continue - let it fall through to regular processing too
+        
         
         # Your existing row processing with enhancements
         for row_idx, row in enumerate(table.rows):