From 81e50cd8340042a38e83917adaf11b1710b63fb5 Mon Sep 17 00:00:00 2001
From: Jan Richter <richter@kajot.cz>
Date: Sun, 14 Jun 2015 12:17:59 +0200
Subject: [PATCH] tukey box plot tests

box plot tests updated according to the new return format of
_box_points() method
---
 pygal/graph/box.py     |  4 +--
 pygal/test/test_box.py | 63 ++++++++++++++++++++++++++++++++++++------
 2 files changed, 56 insertions(+), 11 deletions(-)

diff --git a/pygal/graph/box.py b/pygal/graph/box.py
index 1c47834..e662675 100644
--- a/pygal/graph/box.py
+++ b/pygal/graph/box.py
@@ -212,8 +212,8 @@ class Box(Graph):
             else:  # seq has an odd length
                 return seq[n // 2]
 
-        # sort the copy in case the originals must stay in original order
         outliers = []
+        # sort the copy in case the originals must stay in original order
         s = sorted([x for x in values if x is not None])
         n = len(s)
         if not n:
@@ -252,8 +252,6 @@ class Box(Graph):
                 q0 = s[b0]
                 q4 = s[b4-1]
                 outliers = s[:b0] + s[b4:]
-                #print "Q: [%s,%s,%s,%s,%s] O: %s" \
-                #        % (q0, q1, q2, q3, q4, outliers)
 
             else:
                 q0 = q1 - 1.5 * iqr
diff --git a/pygal/test/test_box.py b/pygal/test/test_box.py
index 492c3a3..95105d6 100644
--- a/pygal/test/test_box.py
+++ b/pygal/test/test_box.py
@@ -22,7 +22,7 @@ from pygal import Box as ghostedBox
 
 def test_quartiles():
     a = [-2.0, 3.0, 4.0, 5.0, 8.0]  # odd test data
-    q0, q1, q2, q3, q4 = Box._box_points(a)
+    (min_s, q0, q1, q2, q3, q4, max_s), outliers = Box._box_points(a)
 
     assert q1 == 7.0 / 4.0
     assert q2 == 4.0
@@ -31,17 +31,17 @@ def test_quartiles():
     assert q4 == 23 / 4.0 + 6.0  # q3 + 1.5 * iqr
 
     b = [1.0, 4.0, 6.0, 8.0]  # even test data
-    q0, q1, q2, q3, q4 = Box._box_points(b)
+    (min_s, q0, q1, q2, q3, q4, max_s), outliers = Box._box_points(b)
 
     assert q2 == 5.0
 
     c = [2.0, None, 4.0, 6.0, None]  # odd with None elements
-    q0, q1, q2, q3, q4 = Box._box_points(c)
+    (min_s, q0, q1, q2, q3, q4, max_s), outliers = Box._box_points(c)
 
     assert q2 == 4.0
 
     d = [4]
-    q0, q1, q2, q3, q4 = Box._box_points(d)
+    (min_s, q0, q1, q2, q3, q4, max_s), outliers = Box._box_points(d)
 
     assert q0 == 4
     assert q1 == 4
@@ -49,9 +49,11 @@ def test_quartiles():
     assert q3 == 4
     assert q4 == 4
 
+
 def test_quartiles_min_extremes():
     a = [-2.0, 3.0, 4.0, 5.0, 8.0]  # odd test data
-    q0, q1, q2, q3, q4 = Box._box_points(a, mode='extremes')
+    (min_s, q0, q1, q2, q3, q4, max_s), outliers = Box._box_points(
+        a, mode='extremes')
 
     assert q1 == 7.0 / 4.0
     assert q2 == 4.0
@@ -60,17 +62,20 @@ def test_quartiles_min_extremes():
     assert q4 == 8.0  # max
 
     b = [1.0, 4.0, 6.0, 8.0]  # even test data
-    q0, q1, q2, q3, q4 = Box._box_points(b, mode='extremes')
+    (min_s, q0, q1, q2, q3, q4, max_s), outliers = Box._box_points(
+        b, mode='extremes')
 
     assert q2 == 5.0
 
     c = [2.0, None, 4.0, 6.0, None]  # odd with None elements
-    q0, q1, q2, q3, q4 = Box._box_points(c, mode='extremes')
+    (min_s, q0, q1, q2, q3, q4, max_s), outliers = Box._box_points(
+        c, mode='extremes')
 
     assert q2 == 4.0
 
     d = [4]
-    q0, q1, q2, q3, q4 = Box._box_points(d, mode='extremes')
+    (min_s, q0, q1, q2, q3, q4, max_s), outliers = Box._box_points(
+        d, mode='extremes')
 
     assert q0 == 4
     assert q1 == 4
@@ -79,6 +84,48 @@ def test_quartiles_min_extremes():
     assert q4 == 4
 
 
+def test_quartiles_tukey():
+    a = []  # empty data
+    (min_s, q0, q1, q2, q3, q4, max_s), outliers = Box._box_points(
+        a, mode='tukey')
+    assert min_s == q0 == q1 == q2 == q3 == q4 == 0
+    assert outliers == []
+
+    # https://en.wikipedia.org/wiki/Quartile example 1
+    b = [6, 7, 15, 36, 39, 40, 41, 42, 43, 47, 49]
+    (min_s, q0, q1, q2, q3, q4, max_s), outliers = Box._box_points(
+        b, mode='tukey')
+    assert min_s == q0 == 6
+    assert q1 == 20.25
+    assert q2 == 40
+    assert q3 == 42.75
+    assert max_s == q4 == 49
+    assert outliers == []
+
+    # previous test with added outlier 75
+    c = [6, 7, 15, 36, 39, 40, 41, 42, 43, 47, 49, 75]
+    (min_s, q0, q1, q2, q3, q4, max_s), outliers = Box._box_points(
+        c, mode='tukey')
+    assert min_s == q0 == 6
+    assert q1 == 25.5
+    assert q2 == (40 + 41) / 2.0
+    assert q3 == 45
+    assert max_s == 75
+    assert outliers == [75]
+
+    # one more outlier, -30
+    c = [6, 7, 15, 36, 39, 40, 41, 42, 43, 47, 49, 75, 77]
+    (min_s, q0, q1, q2, q3, q4, max_s), outliers = Box._box_points(
+        c, mode='tukey')
+    assert min_s == q0 == 6
+    assert q1 == 30.75
+    assert q2 == 41
+    assert q3 == 47.5
+    assert max_s == 77
+    assert 75 in outliers
+    assert 77 in outliers
+
+
 def test_simple_box():
     box = ghostedBox()
     box.add('test1', [-1, 2, 3, 3.1, 3.2, 4, 5])