Fox Presentation Spring 96
CPS 616 January-April 1996 Computational Science Track on Base Technology for the Information Age:
Perl5 and Perl Extensions
Instructor: Geoffrey Fox, wojtek Furmanski
Version 8 April 96
Syracuse University
111 College Place
Syracuse
New York 13244-4100
Abstract of Perl5 and Extensions
See Perl Home Page http://mox.perl.com/perl for background information and resources such as manual!
This Foilset mainly extends the previous Perl4 Presentation with a discussion of some key Perl5 capabilities.
However some features may be advanced Perl4 features
We give an initial Summary of Perl5 Changes and then discuss:
Some old and new functions in Perl
Regular Expression Enhancements
New Syntax especially -> and =>
New subroutine calling and declaration syntax
Hard(address) and Soft(Symbol Table) References
General Data Structures including multidimensional Arrays
Object Oriented Features -- packages, classes and Methods
Some Key Advances in Perl5 -- I
Arbitrary multidimensional Arrays for list [] and Hash {} sectors
Modules allow convenient library structure
Functional but slightly adhoc object oriented class structure hacked onto existing Perl4
C/C++ can be called from Perl and vice versa
tie/untie allow more general database interfaces
AUTOLOAD allows arbitrary action for undefined subroutines
subroutines now can be predeclared before implementation
Some Key Advances in Perl5 -- II
Significant regular expression enhancements
parantheses implemented more cleanly separating backreference ($n) and grouping function
Minimal as well as default maximal greedy matching allowed
Support for comments in regular expression
Full support for pointers (references)
various useful new routines such as my qq my lc uc lcfirst ucfirst
A lot of new pragmas (use english, strict 'vars' 'refs' 'subs' )
Some Advanced Syntax in Perl
New operator => is equivalent to , and use if you feel it looks better
%hash = ( 'key1' , 'value1', 'key2' , 'value2' ); # perl4 is same as
%hash = ( 'key' => 'value1', 'key2' => 'value2' ); # perl5
New Operator -> is dereferencing operator
hard/soft reference to array reference -> [index];
hard/soft reference to hash reference-> {'key'};
class or object -> method;
in hash arrays , quotes are now optional if unambiguous i.e. couldn't be an expression
$days{'Feb'} and $days{Feb} are the same!
Perl Modules -- Packages - I
package fred
$var = 3.14159; # defines fred to be a package and var to be a variable in this package so that variables following this statement should be accessed by
$globalaccess = $fred::var;
Packages can be nested
package fred
....................
package jim
$var = 3.14159; # and now we use syntax
$globalaccess = $fred::jim::var;
Perl Modules -- Packages - II
Modules are packages used as libraries
To reference package HTML::FormatPS
Typically you define a file called
FormatPS.pm -- note .pm NOT .pl -- in directory HTML
Note use of UNIX directory structure and file names to support logical object structure of software -- we saw this quaint (but universal) convention when discussing Java
Example of Module HTML::FormatPS
This (the file FormatPS.pm) starts with following lines
package HTML::FormatPS;
$DEFAULT_PAGESIZE = "A4"; # A Package variable
# Followed by (old) Perl4 pod (plain old documentation) syntax for automatic generation of documentation
=head1 NAME This and Following lines are JUST documentation and ignored by the interpreter
This is Title of documentation
HTML::FormatPS - Format HTML as postscript
# Continued on the next foil ......
Pod Syntax Explaining Use of Module HTML::FormatPS
=head1 SYNOPSIS
require HTML::FormatPS; # This part of documentation defines use of Module
$html = HTML::Parse::parse_htmlfile("test.html"); # access function parse_htmlfile in file Parse.pm in directory HTML
Now define an object of class FormatPS which holds parameters of relevance -- $formatter will hold pointer to new object
$formatter = new HTML::FormatPS
FontFamily => 'Helvetica',
PaperSize => 'Letter'; # Initial Parameters in this Invocation of HTML::FormatPS package
print $formatter->format($html); # run format method in class FormatPS to produce postscript output
An Aside on Perl Pod Notation
The start of pod information is recognized by a command starting with = in column1:
We give HTML approximate equivalents to give intuition!
=head1 heading Roughly equivalent to heading
=head2 heading Roughly equivalent to heading
=over N Indent by N characters
=item text Roughly text
=back Roughly
=cut End Pod Sequence
One uses I to get italic i.e. text in HTML
B for Bold, L for link etc.
See perlpod manual page and look at Perl library code which uses pod notation to generate manual
require and use in Perl5
require Cwd; # Makes notation Cwd:: accessible
$here = Cwd::getcwd(); # correctly accesses function getcwd in module Cwd
$here = getcwd(); # looks for getcwd in current file and probably fails!
On the otherhand:
use Cwd; # Actually imports names (symbol table) from Cwd and
$here = getcwd(); # is equivalent to $here = Cwd::getcwd();
Can use require with Perl prgroms -- not packages
require "fred.pl"; # reads this file and thereby includes any functions included in file. acts like a library call in that will not read fred.pl if already done!
Overview of References in Perl5
Perl5 has a much richer and at times rather confusing syntax for references or pointers as in C or C++
This new feature is used to allow Perl variables to hold "handles" to objects and so implement an object oriented environment.
So in this sense pointers in Perl combine and do not "properly" distinguish classic pointers and objects.
There are hard references and soft or symbol table references
Hard references are new to Perl5 and much more powerful
Soft Symbol Table References - I
One of Perl's "problems" (also its strength if you are knowledgeable) is that one often needs to understand implementation issues to use effectively
Every package has a symbol table (i.e. list of used symbols) called :: so that main symbol table is
%main:: and variable $var in main has symbol table entry $main::{'var'}
*var is equivalent to $main::{'var'}
The symbol $original exists, we can set
*var = *original; # and then $var is another 'name' for $original and @var is separately another name for @original etc.
i.e. $var @var %var have same symbol table entry but will have different hard references
Soft Symbol Table References - II
We can more miraculously set
$name="foo"; # define an innocent ascii string
${$name} = 6; # sets $foo=6 as though $name was a symbolic reference
$$name = 6; # also sets $foo=6
$name->[0] = 4; # sets $foo[0] = 4
${$name x 2} = 6; # sets $foofoo = 6 ; # remember definition of x for strings
@$name = (); # sets @foo to null list while
&$name(arguments); # calls subroutine foo with given arguments!
use strict 'refs'; # FORBIDS symbolic references and above syntax will lead to error messages
*PI =\3.14159; # ensures that $PI is set in a way that you can not override it!
i.e. $PI = 3; # generates an error
Hard References in Perl5 - I
Hard References are more powerful than Typeglobs (symbolic references) and in some cases supercede them
$scalarref = \$foo; # is pointer to a scalar
$getit = $$scalarref; # is same as $getit = $foo
This is called dereferencing or going from pointer(reference) to value
$arrayref = \@array;
$hashref = \%hash;
used as in $$hashref{"key"} = "value";
Hard References in Perl5 - II
$coderef = \&subroutine; # pointer to a subroutine!
accessed by &$coderef(arguments);
$globref = \*STDOUT;
used as in print $globref "We used a pointer to file handle\n";
Note in dereferencing, one can use curlies { } either to disambiguate or to change scalar holding hard reference to a BLOCK returning a reference of correct type
Thus $$scalarref is equivalent to ${$scalarref};
$$hashref{"key"} is equivalent to ${$hashref}{"key"} or $hashref->{"key"}
Anonymous Data Structures and Subroutines
Often one wishes to construct "unnamed" datastructures or subroutines where one keeps track of them by references as opposed to name
This is natural with subroutines which return either a datastructure or subroutine
$arrayref = [1, 2, ['a', 'b', 'c'] ]; # $arrayref is a hard reference to a 2D array with 5 defined elements
$arrayref->[2][1] will give value 'b' as will $$arrayref[2][1]
$arrayref->[0][0] is 1; $arrayref->[0][1] is undef;
$secretsub = sub { print "Support ECS\n" };
executed by &$secretsub;
Data Structures, Arrays of Arrays in Perl5 -I
Read Chapter 4 of Camel Book (second edition) or manual pages on PerlLOL (List of Lists) and PerlDSC (Data Structures Cookbook) -- these have lots of excellent examples
$LoL3D[$x][$y][$z] = scalar func($x,$y,$z); # scalar forces scalar context
is a classic (Fortran like) 3D array except it need NOT be predefined and there are no such things as dimensions and arrays can be ragged with $x=1 having very different $y $z ranges from $x=2 etc
In ragged array, missing elements return undef if accessed
Data Structures, Arrays of Arrays in Perl5 -II
We can also define $indexed2Dhash[$x][$y]{$z} which should of as a hash labelled by two dimensional indices
$hashof2Darray{$x}[$y][$z] should be thought of as a hash whose value is a 2D array
One can freely use such datastructures/arrays as long as use "full" number of indices
Issues which require understanding of implementation occur when you need to manipulate structure "as a whole" with less than full number of indices
There really are only 1D Arrays and Hashs! - I
All multiple dimensional data structures are implemented as arrays of references
In $Lol2D[$x][$y] one stores an array labelled by $x of hard references
Each hard reference is to an anonymous 1D array whose elements are accessed by $y
for $i (1..10) {
@list = somefunc($i); # grab a list labelled by $i
$LoL[$i] = @list; # sets $LoL[$i] as the number of elements in @list as evaluated in scalar context or
$LoL[$i] = scalar @list; # is same as above more pedantically or see next foil
There really are only 1D Arrays and Hashs! - II
continuing from previous foil
$LoL2D[$i] = \@list; # does create a 2D array but \@list is same location each time and so $LoL2D[$i][$j] will give same answer whatever $i chosen with final @list returned being accessed each time
$LoL2D[$i] = [ @list ]; # finally creates intended 2D array as array constructor [ ] will create a fresh reference for each $i
} # End for $i (1..10)
my(@list) = somefunc($i); # my creates a fresh instance each time
$LoL2D[$i] = \@list; # also works but is perhaps less clear
Note my() can occur as here inside any block { } and not just at start of subroutine and defines variables local to block
The -> Pointer Notation -- I
$LoL2D[$i][$j] can be written equivalently $LoL2D[$i]->[$j] but
NOT $LoL2D->[$i]->[$j] or $LoL2D->[$i][$j]
as left hand side of -> MUST be a reference and NOT an array or hash
$ref_to_LoL2D = \@LoL2D; # is allowed and now
access by $ref_to_LoL2D->[$i][$j] or $ref_to_LoL2D->[$i]->[$j]
Note [ .. ] or { .. } create anonymous arrays or hashs respectively which can be assigned to a reference and then dereferenced by ->
( .. ) constructs a list which can be assigned to an Array or Hash
The -> Pointer Notation -- II
@LoL2D = ( [1,2],[1,2,3] ); # Constructs a 2D array
$ref_to_LoL2D = [ [1,2],[1,2,3] ]; # creates a pointer to a 2D array
$arraypt =\@{$LoL2D[$i]}; # extracts a slice ($i'th row) from $LoL2D[$i][$j]
$$arraypt->[$j] is equivalent to $LoL2D[$i][$j]
Perl5 is operationally like Fortran and acts as though right most elements are least significant and stored "consequitively"
If one has defined attributes for students with components such as $student{"grade"}
Then $student[$classmember]{"grade"} is way to write program to address a class of students
Some Remarks about Subroutines - I
Subroutines must typically be predefined (with new sub command) if they are to be accessed with
subname(list); # or one can use
&subname(); # as equivalent to subname()
so that & notation is typically unnecessary
use packagename qw( NAME1 NAME2 NAME3); # imports routines NAME1 NAME2 NAME3 from package packagename
Notice qw() is new Perl5 routine to generate quotes around space separated words
qw(argument); equivalent to split(' ',q(argument));
Some Remarks about Subroutines - II
One can predeclare subroutines with
sub name; # command and use before implementation appears
One can generate subroutines using EVAL or anonymously as in:
sub newprint {
my $x = shift;
return { my $y=shift; print "$x, $y!\n"; }; # return anonymous subroutine
}
$h = newprint("Howdy"); # $h Holds anonymous subroutine
&$h("World"); # Calls anonymous subroutine and we get
Howdy, World! printed
Note the $x in anonymous subroutine is private and retains value Howdy in $h even when you call newprint again with
$g= newprint("Greetings"); # $g has separate instance of $x
Some Remarks about Subroutines - III
Note differences between my() and local();
my($x); # declares $x to be private to this subroutine
local($x); # delares $x to be known to this subroutine and all those routines that it invokes
Typeglob or symbolic reference can be used to pass aruments by reference and not by value
This has usual advantage that subroutine alters "global" and not a "local" copy -- especially relevant for complex datastructures where you do not want expense of copying
Scalars are always passed by reference and not by value and by explicitly changing $_[0..], you can affect global scalars
Some Remarks about Subroutines - IV -- Call by Reference
sub doublearray {
my(*arraypointer) = @_;
foreach $elem (@arraypointer) {
$elem *= 2;
}
} # End routine to double elements of any array
# Suppose @foo and @bar are arrays
doublearray(*foo); # doubles elements of @foo
doublearray(*bar); # doubles elements of @bar
Some Remarks about Subroutines - V -- Separating Arguments
There is a well known problem with ordinary ways of using Perl subroutine arguments
If one has a argument list which is
(@list1, @list2, .... ), then subroutine only sees a single list (array) concatenating the component lists
This can be avoided using hard references with \ operator and shown by example
@tailings = popmany( \@a, \@b, \@c, \@d );
See next foil for code of popmany
Some Remarks about Subroutines - VI -- Separating Arguments
sub popmany { # See previous foil for use
my $aref; # A local scalar to hold pointer to array
my @retlist = (); # An array to hold returned list
foreach $aref ( @_ ) { # $aref loops over arguments
push(@retlist, pop(@$aref) ); # @$aref is global array pointed to in argument
} # Now we have popped off last element in each input array
return @retlist; # @retlist has last element of each array passed
}
Some Remarks about Subroutines - VII -- AUTOLOAD
One can define a default function AUTOLOAD to resolve unsatisfied subroutine references in a gibven (set of) packages.
You set up AUTOLOAD to deal with this case in whatever way you want!
AUTOLOAD is passed arguments that went to called subroutine and name of unsatisfied external is in variable $AUTOLOAD
sub AUTOLOAD { # Call UNIX for unsatisfied externals
my $program = $AUTOLOAD;
$program =~ s/.*:://; # remove any package precursors
system($program, @_);
}
date(); # will now just be executed correctly by above AUTOLOAD
Perl5 Object Model -- I
The object model in Perl5 is not as clear as in Java as the concepts are mixed up with the implementation.
We see same flaws in JavaScript where we "violate" modular programming principles by mixing concept and implementation in the technology which is precisely designed to help programmer keep these separate in his or her own programs!
Objects are references -- not directly variables -- they are typically references returned from subroutines as anonymous datastructures
objects (references) must be "blessed" so they remember what class (module or package) they come from
bless ($self, $class); return $self; # is classic way for a constructor subroutine new to end with $self the datastructure you wish to be stored in object
Perl5 Object Model -- II
Objects are further:
objects are created by a "constructor" which is any suitable function in given module but CONVENTION says for one to call this constructor new
so new is (arbitary) name for a subroutine and not a language construct
A class method is a conventional Perl5 subroutine defined in given class (package) which expects its first argument to be either an object reference or for static methods (independent of object instance) the class name
The class name IS the package name
Constructor for class HTML::FormatPS - I
$formatter = new(HTML::FormatPS , FontFamily => 'Helvetica', PaperSize => 'Letter'); # create instance of class given in first argument with following arguments overriding default paramters
$formatter holds a reference to a blessed hash
remember => is just a comma and arguments to a Perl subroutine are just a single list -- here of 5 entities
In package HTML::FormatPS subroutine new looks like this (continued on next foil)
sub new {
my $class = shift; # set $class as package name and remove from argument list using shift function which takes @_ as default argument in subroutine
Constructor for class HTML::FormatPS - II
# Set up defaults in hash $self which is blessed
my $self = bless {
family => "Times",
mH => mm(40), # mm is local subroutine converting millimetres to points
mW => mm(20),
printpageno => 1,
fontscale => 1,
leading => 0.1,
}, $class; # second argument to bless is class name
$self->papersize($DEFAULT_PAGESIZE); # To be Continued on next foil
Constructor for class HTML::FormatPS - III
# Parse constructor arguments (might override defaults)
while (($key, $val) = splice(@_, 0, 2)) { # get in $key,$value next two elements from @_ and remove them from @_
# Here process key value pair and set $self hash as appropriate
# See original for details which are irrelevant here
}
return($self); # return datastructure
}
splice function in Perl
splice ARRAY,OFFSET,LENGTH,LIST
remove LENGTH elements starting at position OFFSET in ARRAY and replace by elements ( if any) in LIST
shift(@a) is equivalent to splice(@a,0,1)
A Hash of Arrays in class HTML::FormatPS
# We show this example of $PaperSizes{}[] which is a Hash whose key references a 2D array
%PaperSizes =
(
A3 => [mm(297), mm(420)], # mm is built in subroutine
A4 => [mm(210), mm(297)], # to convert millimeteres to points
);
so $PaperSizes{'A4'} returns a two dimensional array
Example of Method in class HTML::FormatPS
$self is of course pointer to datastructure and so following code will alter object passed in first argument
sub papersize {
my($self, $val) = @_; # $self is reference to object
$val = "\u\L$val"; # Uppercase first, lowercase following letters in string $val
my($width, $height) = @{$PaperSizes{$val}};
return 0 unless defined $width;
$self->{papersize} = $val; # reset object attributes
$self->{paperwidth} = $width;
$self->{paperheight} = $height;
1; # return 1 (or 0 if error) BUT not $self
}
Inheritance in Perl5
This is implemented "by hand" using @ISA which is defined for every package and contains list of packages to be searched for unsatisfied externals
package Fred;
require Exporter; # Make package Exporter available to Fred
@ISA = qw(Exporter); # Exporter is to be searched for unsatisfied externals
# See Exporter manual page for more details
Of course AUTOLOAD mechanism kicks in as technique of last resort if cannot find a subroutine anywhere else
Some Predefined Variables in Perl - I
There are the original cryptic two character names and those with a more mnemonic value which are accessible if one invokes
use english; # pragma
Here are a few examples -- we gave some of the predefined variables defining formatted output in first Perl foilset
$ARG or $_ is default name when nothing specified
s/rubbish//; # is equivalent to
$_ =~ s/rubbish/;
chomp(); # is equivalent to
chomp($_); # etc.
Some Predefined Variables in Perl - II
A match m/regexp/; or equivalent s pattern match sets
$MATCH or $& -- The matched string
$PREMATCH or $` to be string before matched string
$POSTMATCH or $' to be string after matched string
$LAST_PAREN_MATCH or $+ contains material in last paranthesis matched
useful when | syntax in regular expression make in unclear which $n is set for last paranthesis
Some functions equivalent to Different Quotes
q(string) or qDstringD for any delimiter D -- interprets string as a literal
This delimiter use is like mDregexpD or sDregexp1Dregexp2D
the one q in q() denotes single quotes
q(string) is equivalent to 'string' except works even if unprotected ' in string
qq(string) or qqDstringD is similarily equivalent to "string" except you do NOT need to protect " inside it
However $variables and \n etc. are interpolated inside string
The two q's in qq() denotes double quotes
qx(string) is similarily equivalent to `string` --
the x in qx stands for execute
Quotemeta() and \Q .. \E Construct
quotemeta("string") protects all regular expression metacharacters
[ab] matches a or b in a regular expression
quotemeta("[ab]") becomes \[ab\] and matches string [ab]
quotemeta("string") is equivalent to "\Qstring\E" where "string" is interpolated before protection
(?..) Constructs in Regular Expressions - I
(?#comment) is a comment in a regular expression
m/[aA][bB](?#matches aB Ab ab AB)/; # is example
/x modifier interprets whitespace in regular expression as for readability and not as "real" characters -- the Perl manual gives an example shown below which removes /* .. */ from C programs
$program =~ s{
/\* # match opening delimiter /*
.*? # minimal match to anything
\*/ # closing delimiter */
}[]gsx; # Replace with nothing (i.e. remove) and specify modifiers g s and x to be operational
(?..) Constructs in Regular Expressions -II
modifier g ensures we match and remove all /* .. */ strings
modifier s treats newlines as part of string
modifier x means that whitespace ignored and # treated like a comment in conventional Perl
Note use of syntax s{old}[new] as equivalent to s%old%new%
One can use any types of parantheses such as s(old){new} etc.
Minimal Matching in Regular Expressions
The default pattern matching in Perl is greedy or maximal size matching
There is now the ? option to designate the selection of match of minimum size
* is replaced by *? to specify minimal match 0 or more times
+? represents minimal match 1 time
?? represents minimal match 0 or 1 times
{n}? minimal match exactly n times
{n,}? minimal match at least n times
{n,m}? minimal match At least n but not more than m times
(?..) Constructs in Regular Expressions -III
(?:regexp) means a simple grouping of regexp as a unit -- equivalent to (regexp) except it does not generate a $n reference
regexp1(?=regexp2) matches to regexp1 followed by regexp2 but regexp2 is not considered part of match
i.e. regexp1 in $MATCH, regexp2 in $POSTMATCH
regexp1(?!regexp2) matches to regexp1 NOT followed by regexp2
(?i) (?m) (?imsx) etc are equivalent to specifying modifiers i, m or i and m and s and x respectively
Some Further Perl4 and Perl5 Functions
-- do and glob!
do "filename.pl"; or more generally do EXPR; # where EXPR returns a string which is taken as a filename
Perl executes contents of file specified by string
This is a good way of loading in a block of subroutines
glob EXPR returns the value of EXPR with conventional UNIX shell filename expansion
glob 'string' is equivalent to where string has UNIX * wildcarding
Upper and Lower case Functions
We have learnt about \L (lower case characters until \E) and \l (lower case next character) and corresponding upper case \U \u so that
"\u\LfOX\E" is "Fox" etc. (\E optional here)
There are a set of function calls implementing these so that
lc(STRING) converts STRING to lower case
lcfirst(STRING) converts first character in STRING to lower case
uc and ucfirst play same role for upper case
ucfirst(lc 'fOX') returns 'Fox'
The defined undef and exists functions
defined(expr); # where expr is typically a variable such as $list[7]
returns true if expr is defined (i.e. not equal to undef)
undef $scalar; undef @list; undef %hash; # set all elements in passed reference to be undefined -- the argument can also be things like $hash{key};
undef on its own represents undefined value for returning from subroutines etc.
undef can be very useful -- for instance you may wish to reuse a hash %parms and execuate undef %parms before re-use.
exists($hash{place}); returns true if place has been defined as a key to %hash-- note this tests existence of associative memory key -- the value $hash{place} may still be undefined!
The map and grep functions
@chars = map( expr , @nums);
Here expr is some Expression or Block which accesses variable $_
map sets $_ to be succesive values in list @nums and returns the successive results of executing BLOCK expr with each value of $_
These results are returned in list context and may give zero one or more entries into total list @chars
A similar construct is grep which acts like map but returns a list containing just the entries in @nums for which expr is TRUE
This is clearly like UNIX grep as
grep ( /regexp/, @listoflines ); # returns just those lines matching regexp
$_ is set by reference to $listoflines[0...] and so altering $_ will alter original @listoflines
pack and unpack Functions -- I
These are generalized tr like functions which convert a list to a string (pack) or string to list/array (unpack) according to a template
We can illustrate with a fragment from a CGI script that reads and interprets data sent to a server in the peculiar application/x-www-form-urlencoded coding scheme
$value =~ tr/+/ /; # Convert + back to blanks
$value =~ /%([a-fA-F0-9][a-fA-F0-9])/pack("C", hex($1))//eg;
pack and unpack Functions -- II
The last line finds encoded hex characters %XY with X,Y in set A-F,a-f,0-9 and replaces them by the "real" character representation
$1 is of course matched XY pair and "C" in pack first argument tells pack to generate character
hex() is a Perl built in function to convert HEX strings to decimal
General syntax is pack(TEMPLATE,LIST) where TEMPLATE is a string of identifiers specifying output style of successive characters
(A is Ascii, l signed long, p pointer, f Float etc.)
ref and scalar functions
ref EXPR returns FALSE unless EXPR is a reference (pointer)
If it is a reference, then ref returns
Package Name if EXPR blessed
REF if a reference
CODE if a subroutine
GLOB if a FileHandle
SCALAR ARRAY HASH if one of three basic types
scalar EXPR forces EXPR to be evaluated in scalar (as opposed to list) context and returns scalar result
We do not need a "list" command as [ ] constructs anonymous arrays and ( ... ) generates a list trivially
tie() and untie() Perl Functions - I
tie() and untie() are described in PerlTIE Man Page
These generate "enchanted" variables ( magical blessing!) which allows one to access what seems to be an ordinary variables in a Perl Program but behind the scenes, the implementation of variable dioes a lot of work which could include computation, data access etc.
Examples in PerlTIE include:
tie a scalar $priority which returns process or user priority by accessing UNIX system
tie a hash to a database so that
$tiedhash{lookupkey} returns value of lookupkey in database
This latter tie is most powerful and could involve SQL access to the database sever when tied hash accessed
tie() and untie() Perl Functions - II
tie VARIABLE, CLASSNAME, LIST
VARIABLE is a scalar, array or hash to be enchanted
Note a given type of tie can have several variables tied which differ by their initial conditions which are specified in LIST which is handed to constructor
The CLASSNAME is a module which must have some special ENCHANTED routines defined -- these are
constructors TIESCALAR() TIEARRAY() or TIEHASH() for three variable types respectively and further functions to define operational access to variables which are listed on following foil
tie() and untie() Perl Functions - III
FETCH -- get variable value
STORE -- store variable value
DESTROY -- destroy variable
Tied hashes also should provide
EXISTS (implements exists)
DELETE (delete one key) CLEAR (clear out all keys)
FIRSTKEY() and together with
NEXTKEY() system implements keys() and each()
Note User provides these functions as they depend on particular behind the scenes manipulations
Note standard classes provide ability to specify user routines BEGIN and END to be run at beginning and end of invocation of a class
Interfacing Perl with C
One can interface Perl5 code with C using an interface constructor called xsubpp
This manipulates an existing C library and allows it to be accessed through a designated Perl module
Simple datatypes can be handled automatically but user must manipulate complex C datastructures in special xsub code
see PerlXS and PerlXSTUT manual entries
Also PerlCALL and PerlEMBED manual pages describe how to call Perl from C
One can add a Perl Interpreter to your C program, execute a Perl statement such as a pattern match or execute a full Perl subroutine