1
00:00:07,060 --> 00:00:08,420
TOMMY: In this video, we'll learn about

2
00:00:08,420 --> 00:00:10,140
redirecting and pipes.

3
00:00:10,140 --> 00:00:12,780
So far, we've been using functions like printf to

4
00:00:12,780 --> 00:00:15,590
output data to the terminal and functions like GetString

5
00:00:15,590 --> 00:00:17,520
to allow the user to provide input to our

6
00:00:17,520 --> 00:00:19,490
program using the keyboard.

7
00:00:19,490 --> 00:00:21,880
Let's quickly take a look at a program that gets a line of

8
00:00:21,880 --> 00:00:25,960
input from the user and then outputs it.

9
00:00:25,960 --> 00:00:28,990
>> On line 7, we're prompting the user for a string, and

10
00:00:28,990 --> 00:00:31,680
then on line 8, we're printing it back out.

11
00:00:31,680 --> 00:00:35,220
Let's compile and run our program.

12
00:00:35,220 --> 00:00:35,900
Great.

13
00:00:35,900 --> 00:00:37,620
The string we provided was echoed back

14
00:00:37,620 --> 00:00:39,170
to us at the terminal.

15
00:00:39,170 --> 00:00:42,110
This happened because the printf function wrote to a

16
00:00:42,110 --> 00:00:46,220
stream called standard out, or s-t-d-out.

17
00:00:46,220 --> 00:00:49,230
When something is written to stdout, it's by default

18
00:00:49,230 --> 00:00:51,110
displayed by the terminal.

19
00:00:51,110 --> 00:00:53,720
>> So that's all well and good, but what if, instead of simply

20
00:00:53,720 --> 00:00:57,700
displaying the string, we wanted to save it to a file?

21
00:00:57,700 --> 00:01:00,470
For example, we might want to remember exactly what our

22
00:01:00,470 --> 00:01:04,450
program did when we gave it a particular input later.

23
00:01:04,450 --> 00:01:07,270
One approach would be to do this in our C program, using

24
00:01:07,270 --> 00:01:09,680
some special functions for writing to files that we'll

25
00:01:09,680 --> 00:01:11,270
see in another video.

26
00:01:11,270 --> 00:01:13,260
Even easier, though, would be to somehow

27
00:01:13,260 --> 00:01:16,090
redirect stdout to a file.

28
00:01:16,090 --> 00:01:19,780
That way, when printf writes to stdout, the contents will

29
00:01:19,780 --> 00:01:21,720
be written to a file rather than

30
00:01:21,720 --> 00:01:23,410
displayed by the terminal.

31
00:01:23,410 --> 00:01:26,690
We can do just that by adding a greater-than sign, followed

32
00:01:26,690 --> 00:01:30,820
by a file name, to the command we use to execute our program.

33
00:01:30,820 --> 00:01:34,730
>> So, rather than simply executing ./redirect, we can

34
00:01:34,730 --> 00:01:38,880
run ./redirect, followed by a greater than sign, followed by

35
00:01:38,880 --> 00:01:41,530
filename, like file.txt.

36
00:01:41,530 --> 00:01:44,290
Let's see what happens.

37
00:01:44,290 --> 00:01:45,130
OK.

38
00:01:45,130 --> 00:01:48,470
Notice that this time, nothing was displayed at the terminal,

39
00:01:48,470 --> 00:01:50,290
but we haven't modified the contents of our

40
00:01:50,290 --> 00:01:52,040
C program at all.

41
00:01:52,040 --> 00:01:56,090
Let's now examine the contents of this directory with ls.

42
00:01:56,090 --> 00:01:56,630
>> All right.

43
00:01:56,630 --> 00:02:00,840
We now have a new file in our directory called file.txt,

44
00:02:00,840 --> 00:02:03,640
which is the file name we supplied when we ran our

45
00:02:03,640 --> 00:02:05,050
Redirect program.

46
00:02:05,050 --> 00:02:08,020
Let's open up file.txt.

47
00:02:08,020 --> 00:02:11,840
And here, we can see that the stdout out of redirect was

48
00:02:11,840 --> 00:02:15,550
written to the file called file.txt.

49
00:02:15,550 --> 00:02:18,470
So let's run the previous command again, but supplying a

50
00:02:18,470 --> 00:02:20,075
different input this time.

51
00:02:25,140 --> 00:02:25,900
Okay.

52
00:02:25,900 --> 00:02:28,205
Let's take a look at file.txt now.

53
00:02:31,070 --> 00:02:34,580
>> We can see here that the file has been overwritten, so our

54
00:02:34,580 --> 00:02:37,120
original input isn't there anymore.

55
00:02:37,120 --> 00:02:40,280
If we instead want to append to this file, putting the new

56
00:02:40,280 --> 00:02:43,600
input below the existing contents of the file, we can

57
00:02:43,600 --> 00:02:46,800
use two greater-than signs instead of just one.

58
00:02:46,800 --> 00:02:48,050
Let's try that.

59
00:02:52,160 --> 00:02:57,910
Now, if we open file.txt again, we can see both of our

60
00:02:57,910 --> 00:02:59,580
input lines.

61
00:02:59,580 --> 00:03:02,180
In some cases, we might want to discard any

62
00:03:02,180 --> 00:03:03,850
output of our program.

63
00:03:03,850 --> 00:03:06,450
Rather than writing the output to a file and then deleting

64
00:03:06,450 --> 00:03:09,310
the file when we're done with it, we can write a to special

65
00:03:09,310 --> 00:03:12,360
file called /dev/null.

66
00:03:12,360 --> 00:03:15,160
When anything is written to /dev/null--

67
00:03:15,160 --> 00:03:16,960
or just devnull for short--

68
00:03:16,960 --> 00:03:18,950
it is automatically discarded.

69
00:03:18,950 --> 00:03:23,290
So think of devnull as a black hole for your data.

70
00:03:23,290 --> 00:03:26,070
>> So now we've seen how the greater than sign can redirect

71
00:03:26,070 --> 00:03:29,610
stdout, let's see how we can redirect standard in--

72
00:03:29,610 --> 00:03:31,250
or s-t-d-in--

73
00:03:31,250 --> 00:03:33,550
the analogue of stdout.

74
00:03:33,550 --> 00:03:36,010
While functions like printf write to the stream called

75
00:03:36,010 --> 00:03:40,500
stdout, GetString and similar functions read from the stream

76
00:03:40,500 --> 00:03:43,770
called stdin, which, by default, is the stream of

77
00:03:43,770 --> 00:03:46,290
characters typed at the keyboard.

78
00:03:46,290 --> 00:03:50,010
We can redirect stdin using the less than sign, followed

79
00:03:50,010 --> 00:03:51,370
by a filename.

80
00:03:51,370 --> 00:03:54,000
Now, rather than prompting the user for input at the

81
00:03:54,000 --> 00:03:57,870
terminal, a program will open the file we specified and use

82
00:03:57,870 --> 00:03:59,790
its lines as input.

83
00:03:59,790 --> 00:04:02,620
>> Let's see what happens.

84
00:04:02,620 --> 00:04:03,280
Great.

85
00:04:03,280 --> 00:04:07,590
The first line of file.txt has been printed to the terminal

86
00:04:07,590 --> 00:04:10,160
because we're calling GetString once.

87
00:04:10,160 --> 00:04:13,170
If we had another call to GetString in our program, the

88
00:04:13,170 --> 00:04:16,149
next line of file.txt would have been used as

89
00:04:16,149 --> 00:04:17,990
input to that call.

90
00:04:17,990 --> 00:04:21,050
Again, we haven't modified our C program at all.

91
00:04:21,050 --> 00:04:23,620
We're only changing how we run it.

92
00:04:23,620 --> 00:04:27,080
And also remember, we haven't redirected stdout this time,

93
00:04:27,080 --> 00:04:28,970
so the output of the program was still

94
00:04:28,970 --> 00:04:31,040
displayed at the terminal.

95
00:04:31,040 --> 00:04:33,500
We can, of course, redirect both stdin

96
00:04:33,500 --> 00:04:37,320
and stdout like this.

97
00:04:37,320 --> 00:04:43,550
Now, file2.txt contains the first line of file.txt.

98
00:04:43,550 --> 00:04:46,140
>> So, using these operators, we've been able to read and

99
00:04:46,140 --> 00:04:48,130
write from text files.

100
00:04:48,130 --> 00:04:51,890
Now, let's see how we can use the output of one program as

101
00:04:51,890 --> 00:04:54,710
the input to another program.

102
00:04:54,710 --> 00:04:56,650
So here's another simple C program I

103
00:04:56,650 --> 00:05:00,190
have here called hello.c.

104
00:05:00,190 --> 00:05:02,617
As you can see, this simply outputs "Hi

105
00:05:02,617 --> 00:05:04,430
there!" to the user.

106
00:05:04,430 --> 00:05:08,890
If I want redirect to use as input the output of hello--

107
00:05:08,890 --> 00:05:10,190
another program--

108
00:05:10,190 --> 00:05:13,920
I could first redirect the stdout of hello to a file called

109
00:05:13,920 --> 00:05:18,960
input.txt, then redirect the stdin of redirect to that same

110
00:05:18,960 --> 00:05:21,190
file--input.txt.

111
00:05:21,190 --> 00:05:26,730
So I can do ./hello > input.txt.

112
00:05:26,730 --> 00:05:28,810
Press Enter to execute this.

113
00:05:28,810 --> 00:05:31,910
Followed by ./redirect <

114
00:05:31,910 --> 00:05:35,270
input.txt, and execute that.

115
00:05:35,270 --> 00:05:38,290
So we can shorten this a bit with a semicolon, which allows

116
00:05:38,290 --> 00:05:41,360
us to run two or more commands on the same line.

117
00:05:41,360 --> 00:05:47,920
So I can say, ./hello > input.txt, semicolon,

118
00:05:47,920 --> 00:05:50,580
./redirect < input.txt.

119
00:05:53,850 --> 00:05:56,740
>> So this works, but it still feels pretty inelegant.

120
00:05:56,740 --> 00:05:59,530
I mean, do we really need this intermediary text file that's

121
00:05:59,530 --> 00:06:02,520
no longer necessary after redirect runs?

122
00:06:02,520 --> 00:06:05,780
Luckily, we can avoid this extra text file using what's

123
00:06:05,780 --> 00:06:07,220
called a pipe.

124
00:06:07,220 --> 00:06:13,740
If I say, ./hello | ./redirect, then the stdout of

125
00:06:13,740 --> 00:06:15,310
the program on the left--

126
00:06:15,310 --> 00:06:16,740
in this case, hello--

127
00:06:16,740 --> 00:06:18,970
will be used as the standard input for the

128
00:06:18,970 --> 00:06:20,370
program on the right.

129
00:06:20,370 --> 00:06:24,850
In this case, redirect. So let's run this.

130
00:06:24,850 --> 00:06:25,930
>> There we go.

131
00:06:25,930 --> 00:06:30,080
We can see that the output of hello was used as the input

132
00:06:30,080 --> 00:06:31,520
for redirect.

133
00:06:31,520 --> 00:06:34,890
By stringing commands together using pipes, we form what's

134
00:06:34,890 --> 00:06:38,120
called a pipeline, since our output is essentially moving

135
00:06:38,120 --> 00:06:40,590
through a sequence of commands.

136
00:06:40,590 --> 00:06:43,570
Using pipes, we can do some cool stuff without needing to

137
00:06:43,570 --> 00:06:45,870
write any code at all.

138
00:06:45,870 --> 00:06:48,760
For example, let's say we want to know how many files are

139
00:06:48,760 --> 00:06:50,630
inside of this directory.

140
00:06:50,630 --> 00:06:55,200
Using a pipe, we can combine the ls command with the wc--

141
00:06:55,200 --> 00:06:56,460
or wordcount--

142
00:06:56,460 --> 00:06:57,850
command.

143
00:06:57,850 --> 00:07:02,230
Ls will output each file in the directory to stdout, and

144
00:07:02,230 --> 00:07:08,040
wc will tell us how many lines were given to it via stdin.

145
00:07:08,040 --> 00:07:12,440
So, if we say ls | wc -l--

146
00:07:12,440 --> 00:07:16,800
supplying the -l flag to wc to tell it to count lines--

147
00:07:16,800 --> 00:07:19,260
we can see exactly how many files are

148
00:07:19,260 --> 00:07:21,940
in the current directory.

149
00:07:21,940 --> 00:07:24,570
>> So let's take a look at one more example.

150
00:07:24,570 --> 00:07:27,740
I have here a file called students.txt,

151
00:07:27,740 --> 00:07:29,600
with a list of names.

152
00:07:29,600 --> 00:07:32,730
However, these names aren't in any order it all, and it looks

153
00:07:32,730 --> 00:07:34,850
like a few names are repeated.

154
00:07:34,850 --> 00:07:38,510
What we want is a list of unique names in alphabetical

155
00:07:38,510 --> 00:07:42,550
order, saved to a file called final.txt.

156
00:07:42,550 --> 00:07:45,210
We could, of course, write a C program to do this for us.

157
00:07:45,210 --> 00:07:46,560
But it's going to get unnecessarily

158
00:07:46,560 --> 00:07:48,560
complex pretty quickly.

159
00:07:48,560 --> 00:07:51,740
Let's instead use pipes and some built-in-tools to solve

160
00:07:51,740 --> 00:07:53,300
this problem.

161
00:07:53,300 --> 00:07:57,760
>> The first thing we need to do is read the file students.txt.

162
00:07:57,760 --> 00:08:00,530
The cat command will do just that.

163
00:08:00,530 --> 00:08:03,230
It will read in the specified file and write

164
00:08:03,230 --> 00:08:05,750
its contents to stdout.

165
00:08:05,750 --> 00:08:07,570
After we've read the text file, we'll

166
00:08:07,570 --> 00:08:09,490
want to sort the names.

167
00:08:09,490 --> 00:08:12,510
The sort command can handle this for us.

168
00:08:12,510 --> 00:08:16,830
Sort will output the line supplied via stdin to stdout

169
00:08:16,830 --> 00:08:19,310
in sorted order.

170
00:08:19,310 --> 00:08:23,450
In order to supply the contents of students.txt to

171
00:08:23,450 --> 00:08:29,600
sort's stdin, we could use a pipe to combine cat and sort.

172
00:08:29,600 --> 00:08:34,440
So I can execute cat students.txt | sort and

173
00:08:34,440 --> 00:08:35,640
press Enter.

174
00:08:35,640 --> 00:08:39,309
And now, we see the contents of students.txt in

175
00:08:39,309 --> 00:08:40,909
alphabetical order.

176
00:08:40,909 --> 00:08:42,860
>> So let's add another command--

177
00:08:42,860 --> 00:08:44,730
uniq, or unique--

178
00:08:44,730 --> 00:08:46,230
to our pipeline.

179
00:08:46,230 --> 00:08:49,810
As you might guess, uniq, when supplied a sorted sequence of

180
00:08:49,810 --> 00:08:53,650
lines via stdin, will output the unique lines.

181
00:08:53,650 --> 00:08:56,910
So now we have cat students.txt

182
00:08:56,910 --> 00:09:00,040
| sort | uniq.

183
00:09:00,040 --> 00:09:03,330
Finally, we can save the output of the pipeline to a

184
00:09:03,330 --> 00:09:09,090
file via cat students.txt | sort | uniq

185
00:09:09,090 --> 00:09:12,440
> final.txt.

186
00:09:12,440 --> 00:09:16,260
So, if we open up final.txt, we have exactly what we were

187
00:09:16,260 --> 00:09:17,270
looking for:

188
00:09:17,270 --> 00:09:20,180
a list of unique names in alphabetical order,

189
00:09:20,180 --> 00:09:22,150
saved to a text file.

190
00:09:22,150 --> 00:09:26,020
By the way, we also could have said sort <

191
00:09:26,020 --> 00:09:32,290
students.txt | uniq > final.txt to do exactly

192
00:09:32,290 --> 00:09:35,400
the same thing, using each of the operators we've seen in

193
00:09:35,400 --> 00:09:36,580
this video.

194
00:09:36,580 --> 00:09:39,540
>> My name is Tommy, and this is CS50.