Read Excel file with multiple headers using RODBC

Question

I have an Excel file which was very beautifully doctored up from which I am extracting some data. There are two layers of "column heads", for instance:

Date
Month Day Year
12    5   2012
12    6   2012
12    12   2012

Where month/day/year are all nested under date. The subheadings are clear enough that I don't need the nominal top row header. However, the bottom row is useful for identifying variables. Furthermore, when I read these using RODBC connections, to my chagrin I realize that the second row is treated as a data element and inconsistent with the labeling in the XLS schema. Any character variable has a header come through accurately, but numeric valued columns return missing values for the second header, regardless of how many rows I return from the query.

> X <- sqlFetch(my.xls, 'this table')
> head(X)

Date F2 F3
NA   NA NA
12   5  2012
12   6  2012
12   12 2012

> X <- sqlQuery(my.xls, 'select top 1 * from "this table$"')
> X
NA NA NA

How can I get these literal values out of the second header?

Answer 1

Package XLConnect was able to extract the pertinent information. I think the problem here is that Window's ODBC drivers can't handle some of the ad-hoc flexibility from Excel such as enforcing column wide formatting of variables. Just a strawman argument for this wonky error.