Merge pull request #33 from wjakethompson/surprise-songs

The Eras Tour Surprise Songs

R/data.R

@@ -156,3 +156,21 @@
 #' @source \url{https://en.wikipedia.org/wiki/Taylor_Swift_albums_discography}
 #' @source \url{https://www.metacritic.com/person/taylor-swift}
 "taylor_albums"
+
+#' The Eras Tour Surprise Songs
+#'
+#' A data set containing all of the surprise songs played on The Eras Tour
+#' through the first North American leg of the tour.
+#'
+#' @format A [tibble][tibble::tibble-package] with `r nrow(eras_tour_surprise)`
+#' rows and `r ncol(eras_tour_surprise)` variables:
+#' * `leg`: The leg of the tour (e.g., North America, South America, etc.).
+#' * `date`: The date of the show in ISO 8601 format (yyyy-mm-dd).
+#' * `city`: The location of the show. For US shows, the location is the city
+#'   and state. For international shows, the location is the city and country.
+#' * `night`: The show number within each city.
+#' * `dress`: The color of the dress Taylor wore on the given night.
+#' * `instrument`: The instrument used to play the song (guitar or piano).
+#' * `song`: The track name of the song.
+#' * `guest`: The special guest (if any) that joined Taylor to play the song.
+"eras_tour_surprise"

data-raw/surprise-songs.R

@@ -0,0 +1,28 @@
+library(tidyverse)
+library(readxl)
+library(here)
+
+devtools::load_all()
+
+eras_tour_surprise <- read_xlsx(here("data-raw", "surprise-songs.xlsx")) %>%
+  mutate(date = as_date(date),
+         night = as.integer(night)) %>%
+  filter(date < today())
+
+# QC data file -----------------------------------------------------------------
+# Check for track names are consistent. Should be 0 rows.
+(bad_name <- eras_tour_surprise %>%
+   filter(!(song %in% taylor_all_songs$track_name)))
+
+# Check that we're using Taylor's Version when possible. Should be 0 rows.
+(not_tv <- eras_tour_surprise %>%
+   left_join(taylor_all_songs %>%
+               filter(is.na(ep) | !ep) %>%
+               distinct(track_name, album_name),
+             join_by(song == track_name),
+             relationship = "many-to-one") %>%
+   filter(!is.na(album_name),
+          !album_name %in% taylor_album_songs$album_name) %>%
+   distinct(song, album_name))
+
+use_data(eras_tour_surprise, overwrite = TRUE)

pkgdown/_pkgdown.yml

@@ -36,6 +36,7 @@ reference:
   - taylor_album_songs
   - taylor_all_songs
   - taylor_albums
+  - eras_tour_surprise
 
 - title: Plotting
 - subtitle: Color palettes

tests/testthat/test-data.R

@@ -10,6 +10,9 @@ test_that("data has expected dimensions", {
   # albums
   expect_equal(ncol(taylor_albums), 5L)
 
+  # surprise songs
+  expect_equal(ncol(eras_tour_surprise), 8L)
+
   albums <-
     unique(taylor_all_songs[which((!taylor_all_songs$ep) &
                                     !is.na(taylor_all_songs$album_name)),
@@ -51,6 +54,11 @@ test_that("column names match documentation expectation", {
   # albums
   expect_equal(colnames(taylor_albums), c("album_name", "ep", "album_release",
                                           "metacritic_score", "user_score"))
+
+  # surprise songs
+  expect_equal(colnames(eras_tour_surprise), c("leg", "date", "city", "night",
+                                               "dress", "instrument", "song",
+                                               "guest"))
 })
 
 test_that("non-TV versions are excluded when possible", {
@@ -66,3 +74,13 @@ test_that("non-TV versions are excluded when possible", {
 
   expect_false(any(exclude_albums %in% taylor_album_songs$album_name))
 })
+
+test_that("surprise songs are the correct version", {
+  expect_true(all(eras_tour_surprise$song %in% taylor_all_songs$track_name))
+
+  no_tv_songs <- eras_tour_surprise$song[grep("Taylor's Version",
+                                              eras_tour_surprise$song,
+                                              invert = TRUE)]
+  expect_false(any(paste(no_tv_songs, "(Taylor's Version)") %in%
+                     taylor_all_songs$track_name))
+})