Formula extensions

NEWS.md

@@ -6,6 +6,16 @@ where the formatting is also better._
 
 ## Development version
 
+### New features
+
+- Support for univariate formulas, e.g., `y ~ 1`, `~ x`, and `~ 0`. These are
+  translated to `x = NULL` or `y = NULL` in the default method call, with
+  automatic type inference: `y ~ 1` (numeric) produces a histogram, `y ~ 1`
+  (factor) produces a barplot, `~ x` (factor) produces a barplot, and `~ x`
+  (numeric) produces a scatterplot against the index. The `~ 0` form is useful
+  for types that don't require x/y, such as `segments` and `rect`. Thanks to
+  @brock for suggestion and discussion. (#534 @zeileis, @grantmcdermott)
+
 ### Aesthetic changes
 
 - The legend plot characters for the `"pointrange"` and `"errorbar"` types now

R/sanitize_type.R

@@ -43,10 +43,23 @@ sanitize_type = function(settings) {
   assert_choice(type, known_types, null.ok = TRUE)
 
   if (is.null(type)) {
-    if (!is.null(x) && (is.factor(x) || is.character(x)) && !(is.factor(y) || is.character(y))) {
+    if (is.null(x) && !(is.factor(y) || is.character(y))) {
+      # enforce histogram type for y ~ 1
+      settings$x = y
+      settings$y = NULL
+      type = type_hist
+    } else if (is.null(x) && (is.factor(y) || is.character(y))) {
+      # enforce barplot type for factor(y) ~ 1
+      settings$x = y
+      settings$y = NULL
+      type = type_barplot
+    } else if ((is.factor(x) || is.character(x)) && is.null(y)) {
+      # enforce barplot type for ~ factor(y)
+      type = type_barplot
+    } else if (!is.null(x) && (is.factor(x) || is.character(x)) && !(is.factor(y) || is.character(y))) {
       # enforce boxplot type for y ~ factor(x)
       type = type_boxplot
-    } else if (is.factor(y) || is.character(y)) {
+    } else if (!is.null(x) && (is.factor(y) || is.character(y))) {
       # enforce spineplot type for factor(y) ~ x
       type = type_spineplot
     } else {

R/sanitize_xylab.R

@@ -37,7 +37,7 @@ sanitize_xylab = function(settings) {
   if (!is.null(ylab)) {
     out_ylab = ylab
   } else if (is_frequency && is.null(y) && !is.null(x)) {
-    out_ylab = "Frequency"
+    out_ylab = if (type == "barplot") "Count" else "Frequency"
   } else if (is_density && is.null(y) && !is.null(x)) {
     out_ylab = "Density"
   } else if (is_ribbon) {

R/tinyformula.R

@@ -74,6 +74,8 @@ tinyframe = function(formula, data, drop = FALSE) {
   ## - formula: (sub-)formula
   ## - data: model.frame from full formula
   if (is.null(formula)) return(NULL)
-  names = sapply(attr(terms(formula), "variables")[-1L], deparse, width.cutoff = 500L)
+  vars = attr(terms(formula), "variables")[-1L]
+  if (is.null(vars)) return(NULL)
+  names = sapply(vars, deparse, width.cutoff = 500L)
   data[, names, drop = drop]
 }

R/tinyplot.R

@@ -1326,13 +1326,16 @@ tinyplot.formula = function(
   m[[1L]] = quote(stats::model.frame)
   mf = eval.parent(m)
 
-  ## extract x
+  ## extract x (if any)
   x = tinyframe(tf$x, mf)
-  xnam = names(x)[[1L]]
-  if (length(names(x)) != 1L) warning(
-    paste("formula should specify exactly one x-variable, using:", xnam),
+  if (!is.null(x)) {
+    xnam = names(x)[[1L]]
+    if (length(names(x)) > 1L) warning(paste("formula should specify at most one x-variable, using:", xnam),
     "\nif you want to use arithmetic operators, make sure to wrap them inside I()")
-  x = x[[xnam]]
+    x = x[[xnam]]
+  } else {
+    xnam = NULL
+  }
 
   ## extract y (if any)
   y = tinyframe(tf$y, mf)
@@ -1341,6 +1344,8 @@ tinyplot.formula = function(
     if (length(names(y)) > 1L) warning(paste("formula should specify at most one y-variable, using:", ynam),
     "\nif you want to use arithmetic operators, make sure to wrap them inside I()")
     y = y[[ynam]]
+  } else {
+    ynam = NULL
   }
 
   ## extract by (if any)
@@ -1369,19 +1374,33 @@ tinyplot.formula = function(
   dens_type = !is.null(type) && (is.atomic(type) && identical(type, "density")) || (!is.atomic(type) && identical(type$name, "density"))
   hist_type = !is.null(type) && (is.atomic(type) && type %in% c("hist", "histogram")) || (!is.atomic(type) && identical(type$name, "histogram"))
   barp_type = !is.null(type) &&  (is.atomic(type) && identical(type, "barplot")) || (!is.atomic(type) && identical(type$name, "barplot"))
-  if (dens_type) {
+  if (is.null(x) && is.null(y)) {
+    # Exception: both x and y NULL (e.g., ~ 0 with type = "segments").
+    # Build labels from xmin/xmax/ymin/ymax names in the original call (m),
+    # since deparse(substitute()) in the default method would see mf[["..."]].
+    if (is.null(xlab) && !is.null(m[["xmin"]]) && !is.null(m[["xmax"]])) {
+      xlab = sprintf("[%s, %s]", deparse1(m[["xmin"]]), deparse1(m[["xmax"]]))
+    }
+    if (is.null(ylab) && !is.null(m[["ymin"]]) && !is.null(m[["ymax"]])) {
+      ylab = sprintf("[%s, %s]", deparse1(m[["ymin"]]), deparse1(m[["ymax"]]))
+    }
+  } else if (is.null(x) && !is.null(y)) {
+    # Exception: univariate y ~ 1 formulas. sanitize_type() will swap x/y and
+    # infer the type (histogram or barplot). Set xlab from the variable name
+    # and let sanitize_xylab() determine ylab after the type is known.
+    if (is.null(xlab)) xlab = ynam
+  } else if (dens_type) {
     # if (is.null(ylab)) ylab = "Density" ## rather assign ylab as part of internal type_density() logic
     if (is.null(xlab)) xlab = xnam
   } else if (hist_type) {
     # if (is.null(ylab)) ylab = "Frequency" ## rather assign ylab as part of internal type_histogram() logic
     if (is.null(xlab)) xlab = xnam
   } else if (is.null(y)) {
-    if (!barp_type) {
+    if (is.factor(x) || is.character(x) || barp_type) {
+      if (is.null(xlab)) xlab = xnam
+    } else {
       if (is.null(ylab)) ylab = xnam
       if (is.null(xlab)) xlab = "Index"
-    } else {
-      if (is.null(ylab)) ylab = "Count"
-      if (is.null(xlab)) xlab = xnam
     }
   } else {
     if (is.null(ylab)) ylab = ynam