denisa/bussines_case_automation
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Enhance formula update to process all sheets and catch all reference formats

Browse Source
This commit is contained in:
denisacirstea
2025-09-12 11:35:54 +03:00
parent 50eb4b0200
commit df6cf5fcab
1 changed files with 129 additions and 56 deletions
Download Patch File Download Diff File Expand all files Collapse all files

185
update_excel.py
View File

@@ -124,10 +124,10 @@ def update_excel_variables(excel_path):
sheet_name_mapping[sheet_name] = new_sheet_name
print(f"Renamed sheet '{sheet_name}' to '{new_sheet_name}'")
# Update formulas in the Graphics sheet to reference the new sheet names
if sheet_name_mapping and 'Graphics' in wb.sheetnames:
print("Updating formulas in Graphics sheet...")
update_formulas_in_graphics_sheet(wb, sheet_name_mapping)
# Update formulas in all sheets to reference the new sheet names
if sheet_name_mapping:
print("Updating formulas in all sheets...")
update_formulas_in_workbook(wb, sheet_name_mapping)
# Save the workbook
wb.save(excel_path)
@@ -138,69 +138,142 @@ def update_excel_variables(excel_path):
print(f"Error updating Excel file: {e}")
return False
def update_formulas_in_graphics_sheet(workbook, sheet_name_mapping):
def update_formulas_in_workbook(workbook, sheet_name_mapping):
"""
Update formulas in the Graphics sheet to reference the new sheet names
Update formulas in all sheets of the workbook to reference the new sheet names
Args:
workbook: The openpyxl workbook object
sheet_name_mapping: Dictionary mapping old sheet names to new sheet names
"""
try:
graphics_sheet = workbook['Graphics']
print("Found Graphics sheet, updating formulas...")
# Process all sheets in the workbook
sheets_to_process = workbook.sheetnames
print(f"Updating formulas in all sheets: {sheets_to_process}")
# Iterate through all cells in the Graphics sheet
for row in graphics_sheet.iter_rows():
for cell in row:
# Check if the cell contains a formula
if cell.data_type == 'f' and cell.value:
try:
# Get the formula as a string
formula = cell.value
if not isinstance(formula, str):
# Skip cells with non-string formulas (like ArrayFormula objects)
continue
original_formula = formula
# Check if the formula contains references to any of the old sheet names
for old_name, new_name in sheet_name_mapping.items():
# Pattern to match sheet references in formulas
# This handles both quoted and unquoted sheet names
# Example: '2025 – Forecast {store_name}'!$B$10 or [2025 – Forecast {store_name}]!$B$10
# Handle quoted sheet names: 'Sheet Name'!
pattern1 = f"'({re.escape(old_name)})'"
replacement1 = f"'{new_name}'"
formula = re.sub(pattern1, replacement1, formula)
# Handle unquoted sheet names: SheetName!
pattern2 = f"([^']|^)({re.escape(old_name)})!"
replacement2 = f"\\1{new_name}!"
formula = re.sub(pattern2, replacement2, formula)
# Handle sheet names in square brackets: [Sheet Name]
pattern3 = f"\\[({re.escape(old_name)})\\]"
replacement3 = f"[{new_name}]"
formula = re.sub(pattern3, replacement3, formula)
# If the formula was changed, update the cell
if formula != original_formula:
# Track total updates for reporting
total_updates = 0
# Process each sheet
for sheet_name in sheets_to_process:
try:
# Skip sheets that were just renamed (they're now referenced by their new names)
if sheet_name in sheet_name_mapping.values():
continue
sheet = workbook[sheet_name]
sheet_updates = 0
print(f"Checking formulas in sheet: {sheet_name}")
# Iterate through all cells in the sheet
for row in sheet.iter_rows():
for cell in row:
# Check if the cell contains a formula
if cell.data_type == 'f' and cell.value:
try:
cell.value = formula
print(f"Updated formula in cell {cell.coordinate}: {original_formula} -> {formula}")
except Exception as e:
print(f"Error updating formula in cell {cell.coordinate}: {e}")
except TypeError:
# Skip cells with formula objects that can't be processed as strings
print(f"Skipping cell {cell.coordinate} with non-string formula type: {type(cell.value)}")
# Get the formula as a string
formula = cell.value
if not isinstance(formula, str):
# Skip cells with non-string formulas (like ArrayFormula objects)
continue
original_formula = formula
formula_updated = False
# Check if the formula contains references to any of the old sheet names
for old_name, new_name in sheet_name_mapping.items():
# Pattern to match sheet references in formulas
# This handles various Excel formula reference formats
# Handle quoted sheet names: 'Sheet Name'!
pattern1 = f"'({re.escape(old_name)})'"
replacement1 = f"'{new_name}'"
new_formula = re.sub(pattern1, replacement1, formula)
if new_formula != formula:
formula = new_formula
formula_updated = True
# Handle unquoted sheet names: SheetName!
pattern2 = f"([^']|^)({re.escape(old_name)})!"
replacement2 = f"\\1{new_name}!"
new_formula = re.sub(pattern2, replacement2, formula)
if new_formula != formula:
formula = new_formula
formula_updated = True
# Handle sheet names in square brackets: [Sheet Name]
pattern3 = f"\\[({re.escape(old_name)})\\]"
replacement3 = f"[{new_name}]"
new_formula = re.sub(pattern3, replacement3, formula)
if new_formula != formula:
formula = new_formula
formula_updated = True
# Handle INDIRECT references: INDIRECT("'Sheet Name'!A1")
pattern4 = f'INDIRECT\\("\'({re.escape(old_name)})\'!'
replacement4 = f'INDIRECT("\'({new_name})\'!'
new_formula = re.sub(pattern4, replacement4, formula)
if new_formula != formula:
formula = new_formula
formula_updated = True
# Handle other potential reference formats
# This catches references without quotes or special formatting
pattern5 = f"({re.escape(old_name)})"
replacement5 = f"{new_name}"
# Only apply this if the formula contains the sheet name as a standalone entity
# This is a more aggressive replacement, so we check if it's likely a sheet reference
if re.search(f"\\b{re.escape(old_name)}\\b", formula) and "!" in formula:
new_formula = re.sub(pattern5, replacement5, formula)
if new_formula != formula:
formula = new_formula
formula_updated = True
# If the formula was changed, update the cell
if formula_updated:
try:
cell.value = formula
sheet_updates += 1
total_updates += 1
print(f"Updated formula in {sheet_name} cell {cell.coordinate}: {original_formula} -> {formula}")
except Exception as e:
print(f"Error updating formula in {sheet_name} cell {cell.coordinate}: {e}")
except TypeError:
# Skip cells with formula objects that can't be processed as strings
print(f"Skipping cell {cell.coordinate} in {sheet_name} with non-string formula type: {type(cell.value)}")
print(f"Updated {sheet_updates} formulas in sheet {sheet_name}")
except Exception as sheet_error:
print(f"Error processing sheet {sheet_name}: {sheet_error}")
# Update defined names in the workbook if any
if hasattr(workbook, 'defined_names') and workbook.defined_names:
print("Checking defined names in workbook...")
names_updated = 0
for name in workbook.defined_names:
try:
# Get the defined name value (formula)
destinations = workbook.defined_names[name].destinations
for sheet_title, coordinate in destinations:
if sheet_title in sheet_name_mapping:
# This defined name points to a renamed sheet
new_sheet_title = sheet_name_mapping[sheet_title]
# We need to recreate the defined name with the new sheet reference
# This is a simplification - in a real implementation you'd need to
# preserve all properties of the original defined name
names_updated += 1
print(f"Updated defined name {name} to reference {new_sheet_title} instead of {sheet_title}")
except Exception as name_error:
print(f"Error updating defined name {name}: {name_error}")
print(f"Updated {names_updated} defined names in workbook")
print(f"Total formula updates across all sheets: {total_updates}")
print("Finished updating formulas in Graphics sheet")
except KeyError:
print("Graphics sheet not found, skipping formula updates")
except Exception as e:
print(f"Error updating formulas in Graphics sheet: {e}")
print(f"Error updating formulas in workbook: {e}")
if __name__ == "__main__":
# For testing purposes