stat_count() respects uniqueness of x

R/position-stack.R

@@ -175,11 +175,10 @@ PositionStack <- ggproto("PositionStack", Position,
       ymax = as.numeric(ifelse(data$ymax == 0, data$ymin, data$ymax))
     )
 
-    data <- remove_missing(
-      data,
-      vars = c("x", "xmin", "xmax", "y"),
-      name = "position_stack"
-    )
+    vars <- intersect(c("x", "xmin", "xmax", "y"), names(data))
+    missing <- detect_missing(data, vars)
+    data[missing, vars] <- NA
+
     flip_data(data, params$flipped_aes)
   },
 

R/stat-count.R

@@ -75,8 +75,7 @@ StatCount <- ggproto("StatCount", Stat,
     x <- data$x
     weight <- data$weight %||% rep(1, length(x))
 
-    count <- as.numeric(tapply(weight, x, sum, na.rm = TRUE))
-    count[is.na(count)] <- 0
+    count <- as.vector(rowsum(weight, x, na.rm = TRUE))
 
     bars <- data_frame0(
       count = count,

tests/testthat/test-geom-col.R

@@ -7,8 +7,8 @@ test_that("geom_col removes columns with parts outside the plot limits", {
     ggplotGrob(p + ylim(0.5, 4)),
     "Removed 3 rows containing missing values or values outside the scale range"
   )
-  expect_warning( # warning created at build stage
-    ggplot_build(p + ylim(0, 2.5)),
+  expect_warning( # warning created at render stage
+    ggplotGrob(p + ylim(0, 2.5)),
     "Removed 1 row containing missing values or values outside the scale range"
   )
 })

tests/testthat/test-scales.R

@@ -111,10 +111,14 @@ test_that("oob affects position values", {
   }
   base + scale_y_continuous(limits = c(-0,5))
 
-  expect_warning(low_censor <- cdata(base + y_scale(c(0, 5), censor)),
+  low_censor <- cdata(base + y_scale(c(0, 5), censor))
+  mid_censor <- cdata(base + y_scale(c(3, 7), censor))
+  handle <- GeomBar$handle_na
+
+  expect_warning(low_censor[[1]] <- handle(low_censor[[1]], list(na.rm = FALSE)),
     "Removed 1 row containing missing values or values outside the scale range")
-  expect_warning(mid_censor <- cdata(base + y_scale(c(3, 7), censor)),
-    "Removed 2 rows containing missing values or values outside the scale range")
+  expect_warning(mid_censor[[1]] <- handle(mid_censor[[1]], list(na.rm = FALSE)),
+    "Removed 3 rows containing missing values or values outside the scale range")
 
   low_squish <- cdata(base + y_scale(c(0, 5), squish))
   mid_squish <- cdata(base + y_scale(c(3, 7), squish))
@@ -127,7 +131,7 @@ test_that("oob affects position values", {
 
   # Bars depend on limits and oob
   expect_equal(low_censor[[1]]$y, c(0.2, 1))
-  expect_equal(mid_censor[[1]]$y, c(0.5))
+  expect_equal(mid_censor[[1]]$y, numeric(0))
   expect_equal(low_squish[[1]]$y, c(0.2, 1, 1))
   expect_equal(mid_squish[[1]]$y, c(0, 0.5, 1))
 })

tests/testthat/test-stat-count.R

@@ -4,3 +4,14 @@ test_that("stat_count() checks the aesthetics", {
   p <- ggplot(mtcars) + stat_count(aes(factor(gear), mpg))
   expect_snapshot_error(ggplot_build(p))
 })
+
+test_that("stat_count() respects uniqueness of `x`", {
+  # For #4609, converting x to factor loses smallest digits, so here we test
+  # if they are retained
+  df <- data_frame0(x = c(1, 2, 1, 2) + rep(c(0, 1.01 * .Machine$double.eps), each = 2))
+  p <- ggplot(df, aes(x)) + stat_count(position = "identity")
+  data <- layer_data(p)
+
+  expect_length(vec_unique(df$x), 4)
+  expect_equal(data$y, rep(1, 4))
+})