Prevent the prior calculated results from being misused

[yutannihilation]

Fix #143

In the following dplyr::transmute(), a column name x in y = x refers to the previously generated column x, which is actually other_col.

data <- data.frame(x = 1, y = 2, other_col = 3)

dplyr::transmute(data,
  x = other_col,
  y = x
)

To ensure the variable refers to the original column, the expressions needs to be wrapped by data.frame and let it get auto-unspliced.

library(patchwork)
devtools::load_all("~/repo/gghighlight/")
#> ℹ Loading gghighlight
#> Loading required package: ggplot2

set.seed(22)
df <- tibble::tibble(
  week = rep(1:3, each = 3),
  values = runif(9),
  x = values,
  id = rep(c("a", "b", "c"), times = 3)
)

p1 <- ggplot(df, aes(week, values, colour = id)) +
  geom_line() +
  gghighlight(id == "a")
#> Warning: Tried to calculate with group_by(), but the calculation failed.
#> Falling back to ungrouped filter operation...
#> label_key: id

p2 <- ggplot(df, aes(week, x, colour = id)) +
  geom_line() +
  gghighlight(id == "a")
#> Warning: Tried to calculate with group_by(), but the calculation failed.
#> Falling back to ungrouped filter operation...
#> label_key: id

p1 * p2

^{Created on 2021-06-05 by the reprex package (v2.0.0)}