The Zen of Perl

Greg J. Badros

badros@cs.washington.edu

University of Washington, Seattle

CSE-341, Summer 1999

(C) 1999, Greg J. Badros—All Rights Reserved

Perl

• is an imperative language

• supports many programming styles(including object-oriented)

• is portable across platforms

“A language for getting your job done!”—Larry WallDesigner and primary implementor of Perl

Design philosophy

• No a priori design, no committees!

• A mix-and-match accumulation of useful features desired by real programmers over many years

• Originally for text processing, generating reports

• Has features from C, Java, Unix shells, awk, and sed

What is it used for?

• Text processing, generating reports

• GUI front-ends to command-line commands

• Systems integration programming

• Web CGI scripting

• …and lots lots more!

Perl is incredibly useful

• Of about 480 of mygeneral-purpose scripts
  • ~ 220 are sh shell scripts(many of these use Perl inside!)
  • ~ 130 are zsh shell scripts
  • ~ 110 are Perl scripts

Perl language features

• Dynamically typed

• Lexical and dynamic scoping

• First-class functions

• Built-in arrays, lists, hash-tables,“regular expressions”

• Module system

• Automatic memory reclamation(via reference counting)

Sample task

Print a report of the users of a given computer system using /etc/passwd

• Input: /etc/passwd file

• Output: Human readable report

• Think about how you would do this in C++ or Java...

Running Perl code

• Can use shebang (sharp-bang) lines when running under Unix varieties:#!/usr/bin/perlmeans to use the binary “/usr/bin/perl” to interpret the remaining lines of the file

• Can also run Perl directly:perl passwd-report

Ultra-fast byte-compilation

• Perl seems to be interpreted—very fast turnaround time

• Actually, it byte-compiles the source code very quickly, saves the byte-codes in memory, and then has a virtual machine that runs those byte codes

• Fast compilation + surprisingly fast execution = easy and quick development

Perl philosophies

• There’s more than one way to do it(TMTOWTDI)

• The long term lazy wayDo it right, since you’ll end up using it over and over again

• 3 great virtues of a programmer:Laziness, impatience, and hubris

Conditionals in Perl

if ($nLines < 0) { $nLines = 0;}

or

$nLines = 0 if $nLines < 0;

Focus on the conditional?

or

Focus on the assignment?

Larry Wall is a linguist

If you make a cup of tea,

I’ll drink it.

or

I’ll drink a cup of tea if you make it.

Variables

• $scalar number, string, reference

• @array heterogeneous

• %hash maps keys to values

• &subroutine usually omit the &

Comparisons

• < == > for comparing numbers

• lt eq gt for comparing strings

"a" < "b" ? undef

"a" lt "b" ? 1

"11" < "2" ? undef

"11" lt "2" ? 1

Interpreted as

FALSE

Interpretedas TRUE

Strings and numbersare one and the same

"11" < 2 ? undef

"11" lt 2 ? 1

• Instead of giving a type error,Perl is defined to give a reasonable meaning to virtually any expression!

• Downside: sometimes the meaning may surprise or confuse you!

Variable interpolationand string literals

my $a = 2;my $b = "World";print STDOUT "Hello $b\n1+1=$a\n";print STDOUT ’Hello $b\n1+1=$a\n’;

Output:Hello World1+1=2

Hello $b\n1+1=$a\n

Variables substitutedfor values inside double quotes

Single quotes result in a stringwith the exact contents

Arrays and lists

my @colors = ("red","green","blue");

$colors[0] ? "red"

$colors[1] = "GREEN";

join(",",@colors) ? "red,GREEN,blue"

$ since valueaccessed

is a scalar

@ since here

we are talkingabout the array

as a whole unit

Hash tables

%longday = ( "Sun" => "Sunday", "Mon" => "Monday", "Tue" => "Tuesday",

…

"Sat" => "Saturday",);$longday{"Mon"} ? "Monday"

Not a typo—

trailing comma

is ignored and

makes editing

easier!

Iteration

for my $d (values %longday) {

print $d, "\n";

}

But could just write:

print join("\n",(values %longday)), "\n";

Reading from files

# Print all lines from standard input

# that contain the substring "greg"

while (<>) {

print if /greg/;

}

Lots of magic here—reads from standard input,and assigns to $_

More magic —as if we wrote:print $_ if ($_ =~ m/greg/)

Regular expressions

• Very powerful “wildcard-like” tool

• Simple cases, just matching substrings

"Hi Greg, how are you" =~ m/greg/ ? undef

"Hi Greg, how are you" =~ m/Greg/ ? 1

"Hi Greg, how are you" =~ m/greg/i ? 1

Regular-expression

control flag:

Ignore case!

Regular expressionmeta-characters

"Hi Greg, how are you" =~ m/G./ ? 1

"Hi Greg, how are you" =~ m/G.e/ ? 1

"Hi Greg, how are you" =~ m/G.*e/ ? 1

* Means zero or moreoccurrences

. Matches any character(except newline)

Greedily chose longest matchinstead of: Gre

Literal meta-charactersin regular expressions

"Hi Greg, how are you" =~ m/G\./ ? undef

"Hi Greg, how are you" =~ m/G.\*e/ ? undef

\ Prevents meta-meaning

Usefulness of regularexpressions

• Wrote 11,000 line static analysis tool for better understanding how C programmers used the C pre-processor in real programs

• Used regular expressions pervasively

• For example, to look for #if, #ifdef, or #endif preprocessor directives:m/^\s*#\s*(if(def)?|endif)\s.*$/