iteration performance on rows with zero Columns

[wPatrick]

The following scenario:

I have an Excel file with 3 worksheets.

Worksheet 1 contains approx. 435 rows
Worksheet 2 contains approx. 222 rows
Worksheet 3 contains approx. 17 rows

Structurally, the worksheets are identical, but the dimensions of the worksheets are messed up.

Worksheet 1 Dimensions: A1:AMH435
Worksheet 2 Dimensions: A1:ALZ882
Worksheet 3 Dimensions: A1:AME471

Note: Everything is fine so far, rows were obviously deleted when the worksheets were created. No error at this point.

I now iterate through each row and measure the time it takes to read the columns of a row:

start := time.Now()
row, err := rows.Columns()
elapsed := time.Since(start)
fmt.Printf("%d,%s,%d\n", rowIndex+1, elapsed, len(row))

I now have the following output for worksheet 3, for example:

...
5,68.82µs,9
6,43.795µs,9
7,40.803µs,9
...
16,40.584µs,9
17,40.682µs,9
18,2.853217ms,0
19,2.855862ms,0
...
65,2.922637ms,0
66,2.859574ms,0
...
471,2.872934ms,0
472,2.880524ms,0

You can see that there are 9 columns per row up to row 17. In these rows the execution time is about 40µs. From row 17 to the end of the worksheet dimensions, however, the execution time is 2-3ms per row, even though there are no columns. To me, this feels like a bug, as it is not clear to me why the first 17 rows are processed faster than the subsequent rows without content. I noticed this because the execution time of worksheet 1 is in the ms range, while I need 2-3 seconds for worksheets 2 and 3. Of course i could do a workaround and end the iteration as soon there are empty lines, but that's not possible in every situation.

I am using v2.8.0

[xuri]

Thanks for your issue. Could you show us a complete, standalone example program or reproducible demo? If you open an existing workbook, please provide the file attachment without confidential info.

[wPatrick]

In the coming weeks, I will provide a sample code along with the corresponding Excel files. Currently, I have a new addition to my family that requires a lot of my time. Please keep the topic open as I am still interested in understanding the problem. I debugged some of the xml parsing and managed to isolate the issue to a certain extent. Unfortunately, my understanding isn’t deep enough to truly comprehend what the code is doing.