From c63ee219e690b1a3ce65c5def23d6ad332c046d0 Mon Sep 17 00:00:00 2001
From: Mattriks <Mattriks@users.noreply.github.com>
Date: Thu, 3 Jan 2019 13:33:35 +1100
Subject: [PATCH] Stat_histogram update

---
 docs/src/gallery/geometries.md         | 11 +++----
 src/statistics.jl                      | 45 +++++++-------------------
 test/testscripts/discrete_histogram.jl |  7 ++--
 3 files changed, 22 insertions(+), 41 deletions(-)

diff --git a/docs/src/gallery/geometries.md b/docs/src/gallery/geometries.md
index c40d12994..2ea21d0be 100644
--- a/docs/src/gallery/geometries.md
+++ b/docs/src/gallery/geometries.md
@@ -203,12 +203,11 @@ hstack(p1,p2)
 ```@example
 using Gadfly, RDatasets
 set_default_plot_size(21cm, 16cm)
-p1 = plot(dataset("ggplot2", "diamonds"), x="Price", Geom.histogram)
-p2 = plot(dataset("ggplot2", "diamonds"), x="Price", color="Cut", Geom.histogram)
-p3 = plot(dataset("ggplot2", "diamonds"), x="Price", color="Cut",
-          Geom.histogram(bincount=30))
-p4 = plot(dataset("ggplot2", "diamonds"), x="Price", color="Cut",
-          Geom.histogram(bincount=30, density=true))
+D = dataset("ggplot2","diamonds")
+p1 = plot(D, x="Price", Geom.histogram)
+p2 = plot(D, x="Price", color="Cut", Geom.histogram)
+p3 = plot(D, x="Price", color="Cut", Geom.histogram(bincount=30))
+p4 = plot(D, x="Price", color="Cut", Geom.histogram(bincount=30, density=true))
 gridstack([p1 p2; p3 p4])
 ```
 
diff --git a/src/statistics.jl b/src/statistics.jl
index e656119da..c633af00b 100644
--- a/src/statistics.jl
+++ b/src/statistics.jl
@@ -332,16 +332,11 @@ function apply_statistic(stat::HistogramStatistic,
                         vals, stat.minbincount, stat.maxbincount)
         end
 
-        if stat.density
-            x_min = Gadfly.concrete_minimum(vals)
-            span = x_max - x_min
-            binwidth = span / d
-            bincounts = bincounts ./ (sum(bincounts) * binwidth)
-        end
-
         binwidth = (x_max - x_min) / d
     end
 
+    stat.density && (bincounts = bincounts ./ (sum(bincounts) * binwidth))
+
     if aes.color === nothing
         T = typeof(x_min + 1*binwidth)
         setfield!(aes, othervar, Array{Float64}(undef, d))
@@ -355,21 +350,15 @@ function apply_statistic(stat::HistogramStatistic,
             getfield(aes, othervar)[j] = bincounts[j]
         end
     else
-        groups = Dict()
+        CT, VT = eltype(aes.color), eltype(vals)
+        groups = Dict{CT,Vector{VT}}(c=>VT[] for c in unique(aes.color))
         for (x, c) in zip(vals, cycle(aes.color))
-            Gadfly.isconcrete(x) || continue
-
-            if !haskey(groups, c)
-                groups[c] = Float64[x]
-            else
-                push!(groups[c], x)
-            end
+            Gadfly.isconcrete(x) && push!(groups[c], x)
         end
         T = typeof(x_min + 1*binwidth)
         setfield!(aes, minvar, Array{T}(undef, d * length(groups)))
         setfield!(aes, maxvar, Array{T}(undef, d * length(groups)))
         setfield!(aes, var, Array{T}(undef, d * length(groups)))
-
         setfield!(aes, othervar, Array{Float64}(undef, d * length(groups)))
         colors = Array{RGB{Float32}}(undef, d * length(groups))
 
@@ -378,7 +367,6 @@ function apply_statistic(stat::HistogramStatistic,
         for (i, (c, xs)) in enumerate(groups)
             fill!(bincounts, 0)
             for x in xs
-                Gadfly.isconcrete(x) || continue
                 if isdiscrete
                     bincounts[round(Int,x)] += 1
                 else
@@ -394,22 +382,13 @@ function apply_statistic(stat::HistogramStatistic,
 
             stack_height += bincounts[1:d]
 
-            if isdiscrete
-                for j in 1:d
-                    idx = (i-1)*d + j
-                    getfield(aes, var)[idx] = j
-                    getfield(aes, othervar)[idx] = bincounts[j]
-                    colors[idx] = c
-                end
-            else
-                for j in 1:d
-                    idx = (i-1)*d + j
-                    getfield(aes, minvar)[idx] = x_min + (j - 1) * binwidth
-                    getfield(aes, maxvar)[idx] = x_min + j * binwidth
-                    getfield(aes, var)[idx] = x_min + (j - 0.5) * binwidth
-                    getfield(aes, othervar)[idx] = bincounts[j]
-                    colors[idx] = c
-                end
+            for j in 1:d
+                idx = (i-1)*d + j
+                getfield(aes, var)[idx] = isdiscrete ? j : x_min + (j - 0.5) * binwidth
+                getfield(aes, minvar)[idx] = x_min + (j - 1) * binwidth
+                getfield(aes, maxvar)[idx] = x_min + j * binwidth
+                getfield(aes, othervar)[idx] = bincounts[j]
+                colors[idx] = c
             end
         end
 
diff --git a/test/testscripts/discrete_histogram.jl b/test/testscripts/discrete_histogram.jl
index 01ed9b2f7..d02144322 100644
--- a/test/testscripts/discrete_histogram.jl
+++ b/test/testscripts/discrete_histogram.jl
@@ -1,5 +1,8 @@
 using Gadfly, Distributions
 
-set_default_plot_size(6inch, 3inch)
+set_default_plot_size(7inch, 3inch)
 
-plot(x=rand(Poisson(20), 1000), Scale.x_discrete, Geom.histogram)
+x = sort!(rand(Poisson(20), 1000))
+p1 = plot(x=x, Scale.x_discrete, Geom.histogram)
+p2 = plot(x=x, Scale.x_discrete, Geom.histogram(density=true))
+hstack(p1, p2)