1
00:00:07,970 --> 00:00:13,635
Welcome back. What if we really want to control the sort order?

2
00:00:13,635 --> 00:00:15,780
Specifying how to break ties,

3
00:00:15,780 --> 00:00:18,565
on the primary property we're using for sorting?

4
00:00:18,565 --> 00:00:23,720
The answer is that, we take advantages of Python's built-in sort order for tuples.

5
00:00:23,720 --> 00:00:26,690
Take a look at this code,

6
00:00:26,690 --> 00:00:31,500
I have a list containing five tuples.

7
00:00:31,580 --> 00:00:34,710
Each tuple has three items in it.

8
00:00:34,710 --> 00:00:37,920
The first tuple has eight, three,

9
00:00:37,920 --> 00:00:39,870
and two, the second tuple has C,

10
00:00:39,870 --> 00:00:42,820
one, and four, and so on.

11
00:00:43,570 --> 00:00:49,110
On line six, I'm going to sort those tuples.

12
00:00:49,110 --> 00:00:51,790
So we're still going to get a list of five tuples,

13
00:00:51,790 --> 00:00:56,755
and then we're just going to print each of them, with an iteration.

14
00:00:56,755 --> 00:01:00,485
What order do you think, they're going to come out in?

15
00:01:00,485 --> 00:01:02,770
Is it just going to do them in

16
00:01:02,770 --> 00:01:06,130
the original order because it doesn't know how to sort tuples?

17
00:01:06,130 --> 00:01:09,160
Is it going to put all of the As first?

18
00:01:09,160 --> 00:01:11,080
So we'll get this one,

19
00:01:11,080 --> 00:01:12,760
and then this one,

20
00:01:12,760 --> 00:01:14,670
and then the B,

21
00:01:14,670 --> 00:01:16,245
and then the two Cs.

22
00:01:16,245 --> 00:01:18,315
Is it going to do some random order?

23
00:01:18,315 --> 00:01:22,180
What's it going to do? Well let's see.

24
00:01:24,560 --> 00:01:27,870
What it's going to do, is put the As first,

25
00:01:27,870 --> 00:01:29,910
and then, the Bs and then the Cs.

26
00:01:29,910 --> 00:01:32,460
So when you sort tuples,

27
00:01:32,460 --> 00:01:38,270
you're really sorting by the first element in those tuples. But there's more.

28
00:01:38,270 --> 00:01:43,860
We have a built-in tie-breaking mechanism, with tuples sorting.

29
00:01:43,860 --> 00:01:47,054
Here, the first values,

30
00:01:47,054 --> 00:01:50,640
were both A, so it went on to the second value,

31
00:01:50,640 --> 00:01:53,370
and two comes before three.

32
00:01:53,370 --> 00:01:57,620
There was only one B, so there was no tie-breaking that needed to happen.

33
00:01:57,620 --> 00:02:00,980
But between the two Cs,

34
00:02:00,980 --> 00:02:03,935
those are equal, so it goes on to the second element.

35
00:02:03,935 --> 00:02:05,225
If those are equal,

36
00:02:05,225 --> 00:02:08,265
it goes on to compare the third elements.

37
00:02:08,265 --> 00:02:10,500
It would even do fourth and fifth as many,

38
00:02:10,500 --> 00:02:12,620
as you had elements in the tuples.

39
00:02:12,620 --> 00:02:16,295
So, when it compares two tuples,

40
00:02:16,295 --> 00:02:19,100
it first compares their first elements.

41
00:02:19,100 --> 00:02:23,075
If one of them is smaller than that whole tuple is smaller,

42
00:02:23,075 --> 00:02:26,960
but if they're equal, it goes on to compare the second elements of the tuples,

43
00:02:26,960 --> 00:02:29,460
and then the third elements and so on.

44
00:02:29,470 --> 00:02:33,410
So that's going to turn out to be useful for us,

45
00:02:33,410 --> 00:02:36,400
when we try to control a sort order for breaking ties,

46
00:02:36,400 --> 00:02:40,835
even when we're not sorting things that are tuples.

47
00:02:40,835 --> 00:02:45,150
Suppose we had really duplicate items,

48
00:02:45,150 --> 00:02:49,600
so we had another A, three, two.

49
00:02:49,600 --> 00:02:52,890
It's exactly the same as the first element.

50
00:02:53,180 --> 00:02:58,800
In that case, it's just going to put both of them in there, one after the other.

51
00:02:59,840 --> 00:03:02,250
So we've got both of the A,

52
00:03:02,250 --> 00:03:03,870
three, two is showing up.

53
00:03:03,870 --> 00:03:05,650
Is never going to collapse them,

54
00:03:05,650 --> 00:03:08,330
if you have six elements to start with and you sort the list,

55
00:03:08,330 --> 00:03:09,920
you're always going to end up with six elements at

56
00:03:09,920 --> 00:03:13,075
the end even if two of them are identical.

57
00:03:13,075 --> 00:03:16,390
All right, so that's sorting tuples.

58
00:03:16,390 --> 00:03:21,820
We're going to take advantage of the Python sort order for tuples,

59
00:03:21,820 --> 00:03:25,954
in order to be able to specify fine-grained control,

60
00:03:25,954 --> 00:03:28,920
on our sort orders for other things,

61
00:03:28,920 --> 00:03:32,160
using tuples, to create a tie-breaking mechanism.

62
00:03:32,160 --> 00:03:34,710
The way we're going to make this tie-breaking mechanism is that,

63
00:03:34,710 --> 00:03:37,080
we're going to make our key function.

64
00:03:37,080 --> 00:03:40,850
As always take one item as input and it's supposed to

65
00:03:40,850 --> 00:03:45,455
return a property of the item that instead of returning one property of the item,

66
00:03:45,455 --> 00:03:47,345
we're going to return a tuple,

67
00:03:47,345 --> 00:03:50,400
containing two properties of the item.

68
00:03:51,080 --> 00:03:53,475
So here's an example.

69
00:03:53,475 --> 00:03:56,925
We've got a list of fruit names,

70
00:03:56,925 --> 00:04:04,400
and we're going to sort them,and the property that we're going to use to sort them,

71
00:04:04,400 --> 00:04:07,920
is defined by this Lambda expression.

72
00:04:08,180 --> 00:04:12,620
It takes as its input one fruit name,

73
00:04:12,620 --> 00:04:17,194
and it returns as its output, a property,

74
00:04:17,194 --> 00:04:21,875
but in this case it's two properties, as a tuple.

75
00:04:21,875 --> 00:04:25,430
The first one is the length of the fruit name,

76
00:04:25,430 --> 00:04:26,780
how long is the word,

77
00:04:26,780 --> 00:04:29,690
and the other is the fruit name itself.

78
00:04:29,690 --> 00:04:32,780
So this is going to produce for peach,

79
00:04:32,780 --> 00:04:38,620
a tuple five comma peach.

80
00:04:40,550 --> 00:04:46,120
For Kiwi, it's going to produce four comma kiwi.

81
00:04:48,320 --> 00:04:52,270
Remember this idea that the key function is sort of,

82
00:04:52,270 --> 00:04:57,205
producing a post-it note that's associated with the item.

83
00:04:57,205 --> 00:04:59,910
So peach has associated with it,

84
00:04:59,910 --> 00:05:04,495
this tuple, and kiwi has associated with it this tuple.

85
00:05:04,495 --> 00:05:08,530
When sorted it's going to decide what order they should go in,

86
00:05:08,530 --> 00:05:12,295
it's sorting them based on these post-it notes, these tuples.

87
00:05:12,295 --> 00:05:16,270
So, for kiwi is going to go before five peach,

88
00:05:16,270 --> 00:05:18,060
because the tuple ordering says,

89
00:05:18,060 --> 00:05:21,085
"Look at the first element of the tuple first".

90
00:05:21,085 --> 00:05:23,950
However, when it comes along,

91
00:05:23,950 --> 00:05:27,380
and sees four comma pair,

92
00:05:28,020 --> 00:05:30,530
it's going to have a tie,

93
00:05:30,530 --> 00:05:32,615
when it compares four with four,

94
00:05:32,615 --> 00:05:35,395
and it's going to then use alphabetic ordering,

95
00:05:35,395 --> 00:05:39,490
as the secondary sort order to break the ties.

96
00:05:39,490 --> 00:05:42,920
So we'll get as our output.

97
00:05:47,990 --> 00:05:53,025
We get the four-letter fruits first; kiwi and pear,

98
00:05:53,025 --> 00:05:56,790
and then the five letter fruits; apple, mango,

99
00:05:56,790 --> 00:06:00,930
and peach, and those are in alphabetic order.

100
00:06:00,930 --> 00:06:05,415
Apple before mango, m before p, mango before peach.

101
00:06:05,415 --> 00:06:08,560
What if we want to have the long words first?

102
00:06:08,560 --> 00:06:12,009
This is just our standard mechanism,

103
00:06:12,009 --> 00:06:13,380
with the sorted function.

104
00:06:13,380 --> 00:06:16,455
We can add the reverse equals True parameter,

105
00:06:16,455 --> 00:06:18,990
and now we'll get blueberry,

106
00:06:18,990 --> 00:06:21,400
to show up first.

107
00:06:21,410 --> 00:06:25,740
That's all fine, except,

108
00:06:25,740 --> 00:06:30,155
it's completely reversed the sort order from what we had before.

109
00:06:30,155 --> 00:06:31,655
So now we have,

110
00:06:31,655 --> 00:06:33,755
peach before mango before apple.

111
00:06:33,755 --> 00:06:38,300
Those are all the five letter words and we now have them in

112
00:06:38,300 --> 00:06:43,980
reverse alphabetic order in addition to reversing the long words to short words.

113
00:06:44,020 --> 00:06:49,505
What if, we wanted to have longest words first,

114
00:06:49,505 --> 00:06:55,070
but break ties, with alphabetic order rather than reverse alphabetical order?

115
00:06:55,070 --> 00:06:57,805
This starts to get pretty tricky.

116
00:06:57,805 --> 00:07:02,360
One solution that's available to us is a little trick.

117
00:07:02,360 --> 00:07:07,625
Instead of using reverse equals true to reverse our sort order,

118
00:07:07,625 --> 00:07:12,680
which will make it so that we reverse both the primary and the secondary property,

119
00:07:12,680 --> 00:07:15,350
I'm going to try to just reverse the primary property.

120
00:07:15,350 --> 00:07:19,235
There is a trick I can use for numeric properties,

121
00:07:19,235 --> 00:07:20,780
like the length of the fruit name.

122
00:07:20,780 --> 00:07:24,504
If I just make all of them be negative values,

123
00:07:24,504 --> 00:07:27,190
so blueberry is now going to be minus nine,

124
00:07:27,190 --> 00:07:29,530
and kiwi is going to be minus four,

125
00:07:29,530 --> 00:07:32,620
minus nine is less than minus four,

126
00:07:32,620 --> 00:07:36,350
and so I'm going to get the longer ones to come first.

127
00:07:36,350 --> 00:07:39,094
But I haven't reversed everything,

128
00:07:39,094 --> 00:07:43,555
so my secondary sort order is still going to be from lowest to highest,

129
00:07:43,555 --> 00:07:45,375
which will be alphabetic.

130
00:07:45,375 --> 00:07:48,030
So I still got blueberry first,

131
00:07:48,030 --> 00:07:53,475
but I now have apple before mango before peach in alphabetic order.

132
00:07:53,475 --> 00:07:57,035
That trick only works if we have a numeric property.

133
00:07:57,035 --> 00:08:01,610
If you had two alphabetic properties and you wanted to do reverse on one,

134
00:08:01,610 --> 00:08:02,640
and not on the other,

135
00:08:02,640 --> 00:08:06,120
it would be harder and I don't have an easy solution for you.

136
00:08:07,030 --> 00:08:09,365
So summarizing.

137
00:08:09,365 --> 00:08:12,470
If you want to specify a tie-breaking property,

138
00:08:12,470 --> 00:08:15,835
have your key function return a tuple.

139
00:08:15,835 --> 00:08:18,330
Like key functions everywhere,

140
00:08:18,330 --> 00:08:21,350
they always take one item from the sequence as input,

141
00:08:21,350 --> 00:08:23,570
but now it's going to return a tuple,

142
00:08:23,570 --> 00:08:27,949
where the first element of the tuple is the primary property to sort by,

143
00:08:27,949 --> 00:08:30,910
the next element is the secondary property to sort by,

144
00:08:30,910 --> 00:08:34,525
and you could even have more elements in the tuple.

145
00:08:34,525 --> 00:08:37,220
We also saw that if you just want to reverse

146
00:08:37,220 --> 00:08:39,710
order for one of the properties but not the others,

147
00:08:39,710 --> 00:08:41,930
instead of using reverse equals true,

148
00:08:41,930 --> 00:08:49,260
you can make the key function return negative of all the numbers. See you next time.