In emptyfield-ds/r_packages_test: Tools for the Avalanche Data Science Team

Your Turn 1

1. Use use_test() to create a new file. Call it "resident_connection"
2. Change the test description (the first argument of test_that()) to "connection is returning valid data"
3. In the test_that() function, remove the default expectations. Replace them with this code:

# `resident_data` is a tibble, isn't empty, and has the right columns
resident_data <- get_resident_data()
expect_s3_class(resident_data, c("tbl_df", "tbl", "data.frame"))
expect_gt(nrow(resident_data), 0)
expect_named(resident_data, c("sector", "residents"))

# `resident_data_dt` is a data.table
resident_data_dt <- get_resident_data(data_table = TRUE)
expect_s3_class(resident_data_dt, c("data.table", "data.frame"))

1. Re-load your package.

2. Press the "Run tests" button in RStudio (above the script pane) or run test_file("tests/testthat/test-resident_connection.R") in the console.

Your Turn 2

1. Run all the tests in the package using test(). Fix the broken tests. Hint: The bug is in R/summarize_data.R
2. Re-run the tests until all of them pass

Your Turn 3

1. Run check() or Cmd/Ctrl + Shift + E
2. Fix the warnings. Re-run check() until you get a clean bill of health. Hint: segment_reactor_output() is in R/summarize_data.R

Your Turn 4

1. Both test-count-donations.R and test-tables.R use donations_test_data. Let's move it to a helper file. First, create and open the file with edit_file("tests/testthat/helper-donations_data.R").

2. Move the code to create donations_test_data into helper-donations_data.R.

3. Remove the donations_test_data code from the two test files.

4. Run the tests.

Take-aways

• testthat automates code testing for you
• Test most functions, but focus on user-facing aspects of the code
• When you find a bug, write a test to make sure you fixed it (and it doesn't show up again!)

emptyfield-ds/r_packages_test documentation built on Jan. 2, 2022, 12:15 a.m.