no skip_rows option for Excel files

[TiburonEl]

Description

I'm having trouble importing an Excel file where the first few rows contain merged cells. I considered skipping these rows during import, but unfortunately, there is no skip_rows option available for Excel files in Polars (only for CSV files).

Is there a way to skip these rows? If not, could this functionality be added in a future release?

[cmdlineluser]

read_options={"skip_rows": ...} will get passed to the underlying engine¹.

Footnotes

1. http://fastexcel.toucantoco.dev/fastexcel.html#ExcelReader.load_sheet_by_name ↩

[TiburonEl]

From what I understand, this parameter can only be an integer, meaning I can skip just one row. What if I want to skip multiple rows? Imagine I was to skip row 1,2 and 5th

[avimallu]

skip_rows is typically intended to ignore a certain number of rows before reading a full table in, because of the Excel file author's formatting preferences.

The operation you're describing is more of a filter, which is what I think you should be using with something like .filter(~pl.int_range(0, pl.len()).is_in(rows_to_skip).

[anapaulagomes]

I'd love to work on this issue.

[anapaulagomes]

This is what worked for me by testing the proposed solutions:

# with read_options
land = pl.read_excel("data/DTB_2022/RELATORIO_DTB_BRASIL_MUNICIPIO.xls", read_options={"header_row": 6})

# without read_options
land = pl.read_excel("data/DTB_2022/RELATORIO_DTB_BRASIL_MUNICIPIO.xls")
land.filter(~pl.int_range(0, pl.len()).lt(5))

I used header_row because it skips the rows and uses the row passed as header.