Linux with Operating System Concepts

literal
characters and 
metacharacters
. The literal characters are characters that must match precisely, in the order 
provided and in the specified form (e.g., an upper-case letter ‘A’ must match an upper-case 
letter ‘A’). The metacharacters are characters that are not compared but instead are applied to 
alter the interpretation of literal characters. For instance, the ‘.’ (period) metacharacter repre-
sents “any character.” The regex 
b.t
will match any string that has the letter ‘b’ followed by 
any single character followed by the letter ‘t.’ The regular expression 
b.t
combines two literal 
characters, ‘b,’ and ‘t,’ and one metacharacter, ‘.’. This regex then will match such strings as
• 
bat
• 
but
• 
bit
• 
b3t
• 
b+t
• 
b.t
• 
b t
The regex will not match the string 
bt
because the . must match a character. In the string 
bt, there is no character between the ‘b’ and ‘t’ to match. The regex 
b.t
will also not match 
b12t
, 
baat,
or 
b%$#!t
because there is not 
exactly
one character between the ‘b’ and ‘t.’
There are several very powerful tools available in Linux that utilize regular expressions. 
Because of this, it is very useful for the Linux user to understand regular expressions and 
be able to utilize them through these tools. For the system administrator, these tools are 
often utilized in a number of ways to simplify their tasks.

---

Regular Expressions
◾
203
Aside from examining regular expressions (and numerous examples), this chapter 
presents three useful tools: 
grep
, 
sed,
and 
awk
. The grep program searches a text file 
for instances of the given regular expression, returning any lines that contain a match-
ing string. This can be very useful for searching multiple files for specific content such as 
people’s names, IP addresses, or program instructions. The egrep program is a variation 
of grep which allows for more metacharacters and so we will primarily examine it when 
we look at grep in Section 6.4. Regular expressions are also used in the programs sed and 
awk although we will see that these programs are far more complex than grep/egrep. We 
examine sed in Section 6.5 and awk in Section 6.6. Regular expressions can also be used 
in both vi and emacs when searching for strings. If you are a programmer, you will also 
find that many modern programming languages have regular expression facilities. One 
of the earliest languages to make use of regex was perl, but we find it in such languages as 
Java, Python, Ruby, PHP, and .Net platform languages (C
++
, C#, J#, ASP, VB, etc).
6.2 METACHARACTERS
In this section, we present the regex metacharacters available in Linux. Each of these will 
be discussed in detail and illustrated through a number of examples. We build upon the 
metacharacters in that we will find some metacharacters can be applied in conjunction. 
That is, some metacharacters not only modify literal characters but can also modify other 
metacharacters. You might find this section to be challenging as this is a complicated topic. 
Working through the examples presented should help.
Table 6.1 lists the various metacharacters that are part of the standard and extended 
regular expression set. In this section, we examine their usage with numerous examples.
The first thing you might notice from the characters in Table 6.1 is that some of them are 
the same as Linux wildcard characters. Unfortunately, the * and ? differ between their use 
as wildcards and their use in regular expressions. So we need to understand the context in 
which the symbol is being used. For instance, the *, when used as part of a regular expres-
sion in grep differs from the use of * when it is used in the ls instruction. We will consider 
this difference in more detail in Section 6.3.
6.2.1 Controlling Repeated Characters through *, 
+
, and ?
The first three metacharacters from Table 6.1 are all used to express a variable number of 
times that the character that precedes the metacharacter can appear. With 
*
, the preceding 
character can appear 0 or more times. With 
+
, the preceding character can appear 1 or 
more times. With 
?
, the preceding character can appear 0 or 1 times exactly.
Consider the following strings:
1. 
aaaabbbbcccc
2. 
abcccc
3. 
accccc
4. 
aaaaaabbbbbb

---

204
◾
Linux with Operating System Concepts
The regular expression 
a*b*c*
would match all four strings because this regular 
expression will match any string that has 0 or more a’s followed by 0 or more b’s fol-
lowed by 0 or more c’s. String #3 has no b and string #4 has no c but * permits this as 
“0 or more.”
The regular expression 
a
+
b
+
c
+
would only match the first two because each letter, a, b, 
and c, must appear at least one time. The regular expression 
a?b?c*
will match the second 
and third strings because the letters a and b must appear 0 or 1 times apiece and the letter 
a appears once in each and the letter b appears once in the second string and 0 times in the 
third string. The first and fourth strings have too many occurrences of a and b to match 
a?b?c*
.
Notice in the previous examples, order must be maintained. The string 
aaaaacccccbbbbbb
would not match 
a
+
b
+
c
+
because the b’s must appear before the 
c’s. Would 
a*b*c*
match this new string? It seems counter-intuitive to say yes it matches 
because the c’s appear before the b’s in the string. But in fact, 
a*b*c*
does match for 
a reason that is not obvious. When we use the * metacharacter, we are saying that the 
string must contain “0 or more” instances of the given literal character. In this case, we are 
requiring that the string contain at least 0 a’s, at least 0 b’s, and at least 0 c’s, in that order. 
It does because the string contains for instance 
aaaaa
which means that the 
a*
portion 
matched. The 
b*
and 
c*
portions also match because immediately following the a’s are 0 
TABLE 6.1 
Regular Expression Metacharacters