Adds ys_prune

[kylebaron]

Summary

• Select any column from a data frame that appears in the spec object
• Column order of the result is same as order in spec
• Optionally report columns in spec that aren't in data
• Error if no columns are in common between data and spec

[andersone1]

@kylebaron

should this test have a data frame with more columns than the spec?

For example

• data has columns: A, B, C
• spec has columns: A, B

The current test has the reverse, and therefore:

names(ys_prune(data, spec)) == names(data)

[kylebaron]

Yes; I was intending fewer columns. Can add more columns too.

[kylebaron]

Added two columns matching nothing.

[andersone1]

@kylebaron

One other question - if I add:

data <- data[, rev(names(data)), drop = FALSE]

after line 12 (data$BAR <- 2) but before line 13 (ans <- ys_prune(data, spec))), the test then fails.

Is this OK?

[kylebaron]

Thanks, @andersone1 ; yes, there was something wrong in the logic for matching / ordering the names.

• Fixed the logic in ys_prune
• Randomized the names in data in the test
• Added expectations for reporting columns not there

Thanks!

[kylebaron]

I reverted that one expectation; I do think we want this to be the check:

  expect_equal(names(ans), names(spec2))

The names in the final answer are identical (including order) to the names in the spec, minus the columns that we dropped.