From e4c4880cffb7662796d0f126aa75c2810cb602eb Mon Sep 17 00:00:00 2001
From: garborg <sean.garborg@gmail.com>
Date: Thu, 6 Feb 2014 19:37:28 -0600
Subject: [PATCH] Add crossjoin

---
 src/DataFrames.jl     |  1 +
 src/dataframe/join.jl | 33 +++++++++++++++++++++++++++++++++
 test/join.jl          | 16 +++++++++++++++-
 3 files changed, 49 insertions(+), 1 deletion(-)

diff --git a/src/DataFrames.jl b/src/DataFrames.jl
index 805d26f638..79749671ff 100644
--- a/src/DataFrames.jl
+++ b/src/DataFrames.jl
@@ -56,6 +56,7 @@ export @~,
        combine,
        complete_cases,
        complete_cases!,
+       crossjoin,
        cut,
        DataFrame,
        DataFrameRow,
diff --git a/src/dataframe/join.jl b/src/dataframe/join.jl
index 90416b1ab1..5d51c3528f 100644
--- a/src/dataframe/join.jl
+++ b/src/dataframe/join.jl
@@ -169,3 +169,36 @@ function Base.join(df1::AbstractDataFrame,
         throw(ArgumentError("Unknown kind of join requested"))
     end
 end
+
+##
+## Crossjoin
+##
+
+let
+    global crossjoin
+    function crossjoin(xs::Union(DataFrame, (Symbol, AbstractVector))...)
+        lens = Int[xlen(x) for x in xs]
+        d = DataFrame()
+        times = 1
+        each = prod(lens)
+        for i in 1:length(xs)
+            ilen = lens[i]
+            each = fld(each, ilen)
+            addx!(d, xs[i], times, each)
+            times *= ilen
+        end
+        d
+    end
+
+    xlen(x::DataFrame) = size(x, 1)
+    xlen(x::(Symbol, AbstractVector)) = size(x[2], 1)
+
+    function addx!(d::DataFrame, x::DataFrame, times::Int, each::Int)
+        for c in x
+            addx!(d, c, times, each)
+        end
+    end
+    function addx!(d::DataFrame, x::(Symbol, AbstractVector), times::Int, each::Int)
+        d[x[1]] = rep(x[2], times, each)
+    end
+end
diff --git a/test/join.jl b/test/join.jl
index 37745eeab0..84bd006ada 100644
--- a/test/join.jl
+++ b/test/join.jl
@@ -32,4 +32,18 @@ module TestJoin
     @test_throws join(df1, df2)
 
     join(df1, df2, on = [:A, :B])
-end
\ No newline at end of file
+
+
+    df = DataFrame(A = 1:2, B = ['b', 'a'])
+    t1 = (:C, [2.4, 3.6])
+    t2 = (:D, [2, 3])
+
+    crossjoin(df, df, t1, t2, df, t2)
+
+    dft = DataFrame(A = [1, 1, 1, 1, 2, 2, 2, 2],
+                    B = ['b', 'b', 'b', 'b', 'a', 'a', 'a', 'a'],
+                    C = [2.4, 2.4, 3.6, 3.6, 2.4, 2.4, 3.6, 3.6],
+                    D = [2, 3, 2, 3, 2, 3, 2, 3])
+
+    @test crossjoin(df, t1, t2) == dft
+end