The Unix Shell: Working with Text¶
Text streams¶
Input and output of most Unix shell programs consists of plain text
streams. Text output from a program can be piped into another program,
or redirected to other streams. The standard streams are stdin (0)
(standard input), stdout (1) (standard output) and stderr (2)
(standard error). The default is to assume that input comes from
stdin and output goes to stdout. We can also stream to and from
a file.
Pipes and redirection¶
Creating a text file from command line¶
Sometimes using a text editor is over-kill. For simple file creation, we can just use re-direction
A single > will create a new file or over-write an existing one.
In [1]:
echo "1 Hello, bash" > hello.txt
In [2]:
ls *txt
hello.txt stderr.txt stdout.txt vietnam.txt
Appending¶
In [3]:
echo "2 Hello, again" >> hello.txt
Special non-printing characters¶
In [4]:
echo -e "3 Hello\n4 again" >> hello.txt
Building a chain of pipes¶
wc -lc reports the number of lines and bytes (usually corresponds to
characters when using English text).
Note that character count is 5 per line and not 4 because cut adds a newline character for each line.
Extract columns 2-5 and then count the number of lines and characters¶
In [7]:
cat hello.txt | cut -c 2-5 | wc -lc
4 20
Capturing error messages¶
The redirection operator > is actually 1> - that is, using stdout. We can also use 2> to redirect the output of stderr. &> means redirect both stdout and stderr, and is useful if for example, you want to direct all output to the same log file for later inspection.
In [9]:
mkdir foo/bar/baz > 'stdout.txt'
mkdir: foo/bar: No such file or directory
As there is nothing from stdout the file is empty¶
In [10]:
cat 'stdout.txt'
We need to use 2> to capture the output from stderr¶
In [11]:
mkdir foo/bar/baz 2> 'stderr.txt'
cat 'stderr.txt'
mkdir: foo/bar: No such file or directory