Handout 11

Overview

• Questions from last time
• Translation subroutine
• Regular expressions

Regular expressions

• This is probably the most well-known aspect of Perl and the most useful for linguists. (If you've had some computational linguistics, it might be useful to know that regular expressions are instances of a regular language and can be expressed with a finite state automaton, but you don't need the technical details to make use of regular expressions.)
• Basically, a regular expression is a restricted way to characterize some set of strings. For example, a Perl regular expression /ab/ characterizes any string where the character a immediately precedes the character b. The regular expression /a[bc]/ characterizes any string with the character sequence ab or ac. Finally, the regular expression /ab*c/ characterizes any string where the character a precedes c with any number of instances of the character b intervening.
• The syntax for checking two strings for a regular expression match is as follows: somestring =~ /regex/. If the regular expression in the slashes (regex) matches somestring, this will evaluate as true (example). Be careful to note that you use the special regular expression operator =~ here, rather than the assignment operator =; this is really easy to mess up.
• You can also make substitutions or replacements using regular expressions with the s/// command. The syntax is as follows: $somevariable =~ s/regex/somestring/. The first instance of the regular expression in $somevariable is replaced with the string given in slashes (example). To replace every single instance of the pattern, we add the g (for "global") flag (example).
• There are lots and lots of special symbols that can be used in Perl regular expressions. (You can see them all with perldoc perlre.) I'll just mention a couple here. First, you can designate any character with a period .. Thus /a..b/ matches any four-character sequence where the first character is a and the last character is b.
• Another set of useful symbols are ^ and $, which match the beginning and end of a string respectively. For example, /^a.*b$/ matches any string where the very first character is a and the very last character is b.
• Finally, with all these special interpretations of characters, we need some way to search for those characters themselves: backslash. For example, /\.\*/ will match the literal character sequence .*.
• To play with regular expressions some, let's look at a very simple perl version of the unix utility grep. (The other special regular expression symbols can be seen with perldoc prelre.)