excel_translator/main.py

from openpyxl import load_workbook, Workbook
import asyncio
from googletrans import Translator
import os
from tqdm import tqdm
import copy
import re
from openpyxl.utils import get_column_letter

async def translate_text(translator, text, target_language):
    try:
        translation = await translator.translate(text, dest=target_language)
        return translation.text
    except Exception as e:
        print(f"Error translating '{text}': {e}")
        return text  # Return the original text if translation fails

def is_formula(text):
    """Check if a cell value is a formula"""
    if isinstance(text, str):
        return text.startswith('=')
    return False

def copy_cell_format(source_cell, target_cell):
    """Copy formatting from source cell to target cell without copying the problematic style index"""
    if source_cell.has_style:
        try:
            # Copy individual style attributes instead of the entire style object
            if source_cell.font:
                target_cell.font = copy.copy(source_cell.font)
            if source_cell.border:
                target_cell.border = copy.copy(source_cell.border)
            if source_cell.fill:
                target_cell.fill = copy.copy(source_cell.fill)
            if source_cell.number_format:
                target_cell.number_format = source_cell.number_format
            if source_cell.protection:
                target_cell.protection = copy.copy(source_cell.protection)
            if source_cell.alignment:
                target_cell.alignment = copy.copy(source_cell.alignment)
            # Copy any hyperlink
            if source_cell.hyperlink:
                target_cell.hyperlink = source_cell.hyperlink
        except Exception as e:
            print(f"Error copying format for cell {source_cell.coordinate}: {e}")

def copy_sheet_properties(source_sheet, target_sheet):
    """Copy all sheet properties from source to target sheet"""
    # Copy column dimensions for all columns
    for col_idx in range(1, source_sheet.max_column + 1):
        col_letter = get_column_letter(col_idx)
        if col_letter in source_sheet.column_dimensions:
            source_dim = source_sheet.column_dimensions[col_letter]
            target_dim = target_sheet.column_dimensions[col_letter]
            
            # Copy all available attributes
            if hasattr(source_dim, 'width') and source_dim.width:
                target_dim.width = source_dim.width
            if hasattr(source_dim, 'hidden'):
                target_dim.hidden = source_dim.hidden
            if hasattr(source_dim, 'outlineLevel'):
                target_dim.outlineLevel = source_dim.outlineLevel
    
    # Copy row dimensions for all rows
    for row_idx in range(1, source_sheet.max_row + 1):
        if row_idx in source_sheet.row_dimensions:
            source_dim = source_sheet.row_dimensions[row_idx]
            target_dim = target_sheet.row_dimensions[row_idx]
            
            if hasattr(source_dim, 'height') and source_dim.height:
                target_dim.height = source_dim.height
            if hasattr(source_dim, 'hidden'):
                target_dim.hidden = source_dim.hidden
            if hasattr(source_dim, 'outlineLevel'):
                target_dim.outlineLevel = source_dim.outlineLevel

async def translate_excel(file_path: str, target_language: str):
    translator = Translator()
    workbook = load_workbook(file_path)
    translated_workbook = Workbook()

    # Count total cells for progress bar
    total_cells = sum(
        sum(1 for _ in sheet.iter_rows())
        for sheet in workbook.worksheets
    )
    
    progress_bar = tqdm(total=total_cells, desc="Translating cells")

    for sheet_name in workbook.sheetnames:
        original_sheet = workbook[sheet_name]
        translated_sheet = translated_workbook.create_sheet(title=sheet_name)
        
        print(f"Processing sheet: {sheet_name} with {original_sheet.max_column} columns")
        
        # Copy sheet properties using the improved function
        copy_sheet_properties(original_sheet, translated_sheet)
        
        # Copy merged cells
        for merged_cell_range in original_sheet.merged_cells:
            translated_sheet.merge_cells(str(merged_cell_range))

        # Process each cell
        for row in original_sheet.iter_rows():
            for cell in row:
                progress_bar.update(1)
                col_idx = cell.column
                row_idx = cell.row
                
                # Create the cell at the same position in the new sheet
                if cell.value is not None:  # Use is not None to include 0 values
                    if is_formula(cell.value):
                        # Don't translate formulas
                        translated_cell = translated_sheet.cell(row=row_idx, column=col_idx, value=cell.value)
                    else:
                        translated_text = await translate_text(translator, str(cell.value), target_language)
                        translated_cell = translated_sheet.cell(row=row_idx, column=col_idx, value=translated_text)
                else:
                    translated_cell = translated_sheet.cell(row=row_idx, column=col_idx)
                
                # Copy formatting
                copy_cell_format(cell, translated_cell)

    # Remove the default sheet created by Workbook
    if "Sheet" in translated_workbook.sheetnames:
        del translated_workbook["Sheet"]

    translated_file_path = os.path.splitext(file_path)[0] + f"_translated_{target_language}.xlsx"
    translated_workbook.save(translated_file_path)
    progress_bar.close()
    print(f"Translated file saved as: {translated_file_path}")

async def main():
    input_file = r"F:\Dev\excel-translator\data\sample\test_sample.xlsx"
    language = "fr"  # French
    await translate_excel(input_file, language)

if __name__ == "__main__":
    asyncio.run(main())