Troubleshooters.Com and Code Corner Present
Steve Litt's PERLs of Wisdom:
PERL Regular Expressions
(With Snippets)
Steve Litt is the author of Troubleshooting Techniques of the Successful Technologist, Rapid Learning: Secret Weapon of the Successful Technologist, and Samba Unleashed. |
Contents
- Introduction
- What They Are
- Doing String Comparisons
- Doing String Selections (parsing)
- Doing Substitutions
- Doing Translations
- Greedy and Ungreedy Matching
- Resolving Doubledots in A Filepath
- Kewl Splitpath One Liner Regex
- Using a Variable as a Match Expression:
- Symbol Explanations:
Introduction
Without regular expressions, PERL would be a fast development environment. Probably a little faster than VB for console apps. With the addition of regular expressions, PERL exceeds other RAD environments five to twenty-fold in the hands of an experienced practitioner, on console apps whose problem domains include parsing (and that's a heck of a lot of them).Regular expressions is a HUGE area of knowledge, bordering on an art. Rather than regurgitate the contents of the PERL documentation or the plethora of PERL books at your local bookstore, this page will attempt to give you the 10% of regular expressions you'll use 90% of the time. Note that for this reason we assume all strings to be single-line strings containing no newline chars.
What They Are
Regular expressions are a syntax, implemented in PERL and certain other environments, making it not only possible but easy to do some of the following:- Complex string comparisons
- $string =~ m/sought_text/; # m before the first slash is the "match" operator.
- Complex string selections
- $string =~ m/whatever(sought_text)whatever2/;
- $soughtText = $1;
- Complex string replacements
- $string =~ tr/originaltext/newtext/; # tr before first slash is "translate" operator.
- Parsing based on the above abilities
Doing String Comparisons
We start with string comparisons because they're the easiest, and yet most of what's contained here is applicable in selecting and replacing text.Simple String Comparisons
The most basic string comparison is$string =~ m/sought_text/;The above returns true if string $string contains substring "sought_text", false otherwise. If you want only those strings where the sought text appears at the very beginning, you could write the following:
$string =~ m/^sought_text/;Similarly, the $ operator indicates "end of string". If you wanted to find out if the sought text was the very last text in the string, you could write this:
$string =~ m/sought_text$/;Now, if you want the comparison to be true only if $string contains the sought text and nothing but the sought text, simply do this:
$string =~ m/^sought_text$/;Now what if you want the comparison to be case insensitive? All you do is add the letter i after the ending delimiter:
$string =~ m/^sought_text$/i;
Using Simple "Wildcards" and "Repetitions"
Calling these "wildcards" may actually conflict with the theoretical grammer and syntax of PERL, but in fact is the most intuitive way to think of it, and will not lead to any coding mistakes.. Match any characterYou can follow any character, wildcard, or series of characters and/or wildcard with a repetiton. Here's where you start getting some power:
/w Match "word" character (alphanumeric plus "_")
/W Match non-word character
/s Match whitespace character
/S Match non-whitespace character
/d Match digit character
/D Match non-digit character
/t Match tab
/n Match newline
/r Match return
/f Match formfeed
/a Match alarm (bell, beep, etc)
/e Match escape
/021 Match octal char ( in this case 21 octal)
/xf0 Match hex char ( in this case f0 hexidecimal)
* Match 0 or more timesNow for some examples:
+ Match 1 or more times
? Match 1 or 0 times
{n} Match exactly n times
{n,} Match at least n times
{n,m} Match at least n but not more than m times
$string =~ m//s*rem/i; #true if the first printable text is rem or REM
$string =~ m/^/S{1,8}/./S{0,3}/; # check for DOS 8.3 filename
# (note a few illegals can sneak thru)
Using Groups ( ) in Matching
Note: Many situations can be done either with groups ( ) or character classes [ ]. Groups are less quirky and they more often yield the results you were looking for.
Groups are regular expression characters surrounded by parentheses. They have two major uses:
- To allow alternative phrases as in /(Clinton|Bush|Reagan)/i. Note that for single character alternatives, you can also use character classes.
- As a means of retrieving selected text in selection, translation and substitution, used with the $1, $2, etc scalers.
Powerful regular expressions can be made with groups At its simplest, you can match either all lowercase or name case like this:
if($string =~ m/(B|b)ill (C|c)linton/)Detect all strings containing vowels
{print "It is Clinton, all right!/n"}
if($string =~ m/(A|E|I|O|U|Y|a|e|i|o|u|y)/)Detect if the line starts with any of the last three presidents:
{print "String contains a vowel!/n"}
if($string =~ m/^(Clinton|Bush|Reagan)/i)Note that the parenthesized element will appear as $1 statements that follow the regular expression. That's OK. If you don't want to use $1, just ignore it. The use of $1, etc, will be explained in the section on Doing String Selections.
{print "$string/n"};
Using Character Classes [ ]
Character classes are alternative single characters within square brackets, and are not to be confused with OOP classes, which are blueprints for objects. If not used carefully, they can yield unexpected results. Remember that groupsare an alternative.Character classes have three main advantages:
- Shorthand notation, as [AEIOUY] instead of (A|E|I|O|U|Y). This advantage is minor at best.
- Character Ranges, such as [A-Z].
- One to one mapping from on class to another, as in tr/[a-z]/[A-Z]. This is essential! It will be discussed in the section on translations.
THE WHOLE THING IN THE SQUARE BRACKETS REPRESENTS EXACTLY ONE CHARACTER!!!
Did I shout loud enough? It may be tempting to do something like this:if($string =~ /[Clinton|Bush|Reagan]/){$office = "President"}The above may even appear to work upon casual testing. Don't do it. Remember that everything inside the brackets represents ONE character, simply listing all it's alternative possibilities.
Other Quirks
I haven't fully investigated this yet, but character classes seem to sometimes do goofy things in regular expressions where the case is ignored (i after the trailing delimiter).Special Characters Inside the Square Brackets
As we've already seen, a hyphen is used to indicate all characters in the colating sequence between the character on the hyphen's left and the character on its right.An uparrow (^) at immediately following the opening square bracket means "Anything but these characters", and effectively negates the character class. For instance, to match anything that is not a vowel, do this:
if($string =~ /[^AEIOUYaeiouy]/){print "This string contains a non-vowel"}Contrast to this:
if($string !~ /[AEIOUYaeiouy]/){print "This string contains no vowels at all"}
Best Uses of Character Classes
Print all people whose name begins with A through Eif($string =~ m/^[A-E]/)If character classes are giving you quirky results, consider using groups!
{print "$string/n"}
Matching: Putting it All Together
Print everyone whose last name is Clinton, Bush or Reagan. Each element of list is first name, blank, last name, and possibly more blanks and more info after the last name. Study this til you understand it.if($string =~ m/^/S+/s+(Clinton|Bush|Reagan)/i)Print every line with a valid phone number.
{print "$string/n"};
if($string =~ m/[/)/s/-]/d{3}-/d{4}[/s/./,/?]/)
{print "Phone line: $string/n"};
Doing String Selections (Parsing)
If regular expressions' only benefit was looking for a (albeit complex) string within a string, it wouldn't be worth learningl. Regular expressions (and PERL itself, for that matter) really start earning their keep by allowing you to select and process substrings based on what they contain, and the context in which they appear.For instance, create a program whose input is a piped in directory command and whose output is stdout, and whose output represents a batch file which copies every file (not directory) older than 12/22/97 to a directory called /oldie. This would be pretty nasty in C or C++. The directory output would look something like this:
Volume in drive D has no labelUUUUgly! I'd hate to do this in C or C++. But wait. It's 18 lines in Perl?
Volume Serial Number is 4547-15E0
Directory of D:/polo/marco
. <DIR> 12-18-97 11:14a .
.. <DIR> 12-18-97 11:14a ..
INDEX HTM 3,237 02-06-98 3:12p index.htm
APPDEV HTM 6,388 12-24-97 5:13p appdev.htm
NORM HTM 5,297 12-24-97 5:13p norm.htm
IMAGES <DIR> 12-18-97 11:14a images
TCBK GIF 532 06-02-97 3:14p tcbk.gif
LSQL HTM 5,027 12-24-97 5:13p lsql.htm
CRASHPRF HTM 11,403 12-24-97 5:13p crashprf.htm
WS_FTP LOG 5,416 12-24-97 5:24p WS_FTP.LOG
FIBB HTM 10,234 12-24-97 5:13p fibb.htm
MEMLEAK HTM 19,736 12-24-97 5:13p memleak.htm
LITTPERL <DIR> 02-06-98 1:58p littperl
9 file(s) 67,270 bytes
4 dir(s) 132,464,640 bytes free
while(<STDIN>)Not bad for 18 lines of code. It could have been shorter, but I wanted to keep it readable. In the snippet above, $1, $2, $3 and $4 are the scalers inside the first, second, third and fourth parenthesis sets. The first three are re-assembled into a yymmdd date string which can be compared with the constant "971222". The fourth holds the filename which will be copied to the /oldie directory if it's not a directory, it's a line with a date, and the date is before 971222. This is the true power of regular expressions and PERL.
{
my($line) = $_;
chomp($line);
if($line !~ /<DIR>/) #directories don't count
{
#** only lines with dates at position 28 and (long) filename at pos 44 **
if ($line =~ /.{28}(/d/d)-(/d/d)-(/d/d).{8}(.+)$/)
{
my($filename) = $4;
my($yymmdd) = "$3$1$2";
if($yymmdd lt "971222")
{
print "copy $filename //oldie/n";
}
}
}
}
Now count the bytes in the directory:
my($totalBytes) = 0;Note the group within a group, where the inner one is used for character alternation, and the outer is used as a selection.
while(<STDIN>)
{
my($line) = $_;
chomp($line);
if($line !~ /<DIR>/) #directories don't count
{
#*** only lines with dates at position 28 ****
if ($line =~ /.{12}((/d| |,){14}) /d/d-/d/d-/d/d/)
{
my($bytes) = $1;
$bytes =~ s/,//; #substitute nothing for comma -- delete commas
$totalBytes += $bytes;
}
}
}
print "$totalBytes bytes in directory./n";
Doing Substitutions
Replace every "Bill Clinton" with an "Al Gore"$string =~ s/Bill Clinton/Al Gore/;Now do it ignoring the case of bIlL ClInToN.
$string =~ s/Bill Clinton/Al Gore/i;
Doing Translations
Translations are like substitutions, except they happen on a letter by letter basis instead of substituting a single phrase for another single phrase. For instance, what if you wanted to make all vowels upper case:$string =~ tr/[a,e,i,o,u,y]/[A,E,I,O,U,Y]/;Change everything to upper case:
$string =~ tr/[a-z]/[A-Z]/;Change everything to lower case
$string =~ tr/[A-Z]/[a-z]/;Change all vowels to numbers to avoid "4 letter words" in a serial number.
$string =~ tr/[A,E,I,O,U,Y]/[1,2,3,4,5]/;
Greedy and Ungreedy Matching
Perl regular expressions normally match the longest string possible. For instance:my($text) = "mississippi";Run the preceding code, and here's what you get:
$text =~ m/(i.*s)/;
print $1 . "/n";
ississIt matches the first i, the last s, and everything in between them. But what if you want to match the first i to the s most closely following it? Use this code:
my($text) = "mississippi";Now look what the code produces:
$text =~ m/(i.*?s)/;
print $1 . "/n";
isClearly, the use of the question mark makes the match ungreedy. But theres another problem in that regular expressions always try to match as early as possible. Read on...
Resolving Doubledots in A Filepath
Doubledots are placefillers for "go up one directory" in a file path. Typically, when you desire to create an absolute path, you want to resolve them by deleting them and the level of directory above them. For instance, /a/b/../whatever becomes /a/whatever.This is MUCH trickier than it might seem. It's likely that all your ideas about greedy matching, replacement strings and the like won't work. Here's the regular expression to resolve A SINGLE double dot:
$text =~ s///[^//]*///././/;In English, this says "find a slash, followed by any number of nonslashes, followed by a slash, followed by two dots, and replace them with nothing. This technique will resolve doubledots in a string as long as that string has only one doubledot. But the plot thickens...
Doubledots can occur alternatively with directories (/a/b/../c/../d) or nested (/a/b/c/../../d). The best way I've found to reliably resolve all doubledots is to make a function that loops through the preceding regular expression until there are no more doubledots. Here's the function:
sub deleteDoubleDots($)The preceding function will resolve all doubledots, be they alternating or nested, or combinations thereof.
{
while($_[0] =~ m//././)
{
$_[0] =~ s///[^//]*///././/;
}
}
Kewl Splitpath One Liner Regex
Check out this splitpath command:my($text) = "/etc/sysconfig/network-scripts/ifcfg-eth0";Is that cool or what?
my($directory, $filename) = $text =~ m/(.*//)(.*)$/;
print "D=$directory, F=$filename/n";
Using a Variable as a Match Expression
You can use a variable inside the match expression. This yields tremendous power. Simply place the variable name between the forward slashes, and the expression will be sought in the string. Here's an example:#!/usr/bin/perl -w |
The preceding code produces the following output.
[slitt@mydesk slitt]$ ./junk.pl |
As you can see, you can seek a regex expression stored in a variable, and you can retrieve the result in $1.
Symbol Explanations:
=~
This operator appears between the string var you are comparing, and the regular expression you're looking for (note that in selection or substitution a regular expression operates on the string var rather than comparing). Here's a simple example:$string =~ m/Bill Clinton/; #return true if var $string contains the name of the president
$string =~ tr/Bill Clinton/Al Gore/; #replace the president with the vice president
!~
Just like =~, except negated. With matching, returns true if it DOESN'T match. I can't imagine what it would do in translates, etc./
This is the usual delimiter for the text part of a regular expression. If the sought-after text contains slashes, it's sometimes easier to use pipe symbols (|) for delimiters, but this is rare. Here are simple examples:$string =~ m/Bill Clinton/; #return true if var $string contains the name of the president
$string =~ tr/Bill Clinton/Al Gore/; #replace the president with the vice president
m
The match operator. Coming before the opening delimiter, this is the "match" operator. It means read the string expression on the left of the =~, and see if any part of it matches the expression within the delimiters following the m. Note that if the delimiters are slashes (which is the normal state of affairs), the m is optional and often not included. Whether it's there or not, it's still a match operation. Here are some examples:$string =~ m/Bill Clinton/; #return true if var $string contains the name of the president
$string =~ /Bill Clinton/; #same result as previous statement
^
This is the "beginning of line" symbol. When used immediately after the starting delimiter, it signifies "at the beginning of the line". For instance:$string =~ m/^Bill Clinton/; #true only when "Bill Clinton" is the first text in the string
$
This is the "end of line" symbol. When used immediately before the ending delimiter, it signifies "at the end of the line". For instance:$string =~ m/Bill Clinton$/; #true only when "Bill Clinton" is the last text in the string
i
This is the "case insensitivity" operator when used immediately after the closing delimiter. For instance:$string =~ m/Bill Clinton/i; #true when $string contains "Bill Clinton" or BilL ClInToN"