merge_range works only one time in constant memory mode

Hi everyone,

im using Python 3.4.3 , xlsxwriter 0.7.3

here a short code example:

workbook = xlsxwriter.Workbook("text.xlsx", {'constant_memory': True})

worksheet = workbook.add_worksheet()

header = workbook.add_format({'bold': True, 'border': 1,
                              'border_color': "#000000",
                              'bg_color': '#808080',
                              'valign': "vcenter",
                              'align': "center",
                              'font_color': '#FFFFFF'})

header.set_text_wrap()

worksheet.merge_range(5,0,6,0, 'Ebene', header)
worksheet.merge_range(5,1,6,1, 'Monat', header)
worksheet.merge_range(5,2,6,2, 'Marke', header)
worksheet.merge_range(5,3,6,3, 'abcde', header)

workbook.close()

and the result shown in Excel 2013

no matter how i reorder it, always the first call of merge_range works properly, the following ones are broken like above.

Remove the {'constant_memory': True} and everything works fine.

regards

[jmcnamara]

Hi,

I’ve added documentation to highlight that merged ranges don’t work when constant_memory mode is enabled.

It is possible to workaround this limitation by writing the merge range cells in row-column order and later calling merge_range(). This requires filling in any non data cells in the range with blank cells. Here is an example based on yours:

import xlsxwriter

workbook = xlsxwriter.Workbook("text2.xlsx", {'constant_memory': True})

worksheet = workbook.add_worksheet()

header = workbook.add_format({'bold': True, 'border': 1,
                              'border_color': "#000000",
                              'bg_color': '#808080',
                              'valign': "vcenter",
                              'align': "center",
                              'font_color': '#FFFFFF'})

header.set_text_wrap()

# Write the merged range cells in row column order so that they conform to the
# rule for 'constant_memory' mode. Fill in any non-text cells with blank
# formatted cells.
worksheet.write(5, 0, 'Ebene', header)
worksheet.write(5, 1, 'Monat', header)
worksheet.write(5, 2, 'Marke', header)
worksheet.write(5, 3, 'abcde', header)

worksheet.write_blank(6, 0, None, header)
worksheet.write_blank(6, 1, None, header)
worksheet.write_blank(6, 2, None, header)
worksheet.write_blank(6, 3, None, header)

# Some other text.
worksheet.write(8, 0, 'Later')

# Right the merge ranges at the end.
worksheet.merge_range(5, 0, 6, 0, 'Ebene', header)
worksheet.merge_range(5, 1, 6, 1, 'Monat', header)
worksheet.merge_range(5, 2, 6, 2, 'Marke', header)
worksheet.merge_range(5, 3, 6, 3, 'abcde', header)

workbook.close()

To fix this properly would require a reasonable amount of work that probably isn’t worth it for the small number of people that this would affect.

So I’m closing this as a documented limitation.

Regards,

John

[yunfan]

@jmcnamara thanks for you time. so the problem is you need to feed the writer iterated, from my view, its like the common issue among SQL ORM libraries. many has a autocommit option which let the user’s operation be send to database, but you still could hold some for your purpose and commit explicit

or you could treat writer job like a transaction in those SQL operations. mostly its single statement, while sometime it could accept multi statement. in this case, mostly you could send single row, but sometimes you could send multi rows with merge operation done