README file for Data::Walker 1.05 NAME Data::Walker - A tool for navigating through Perl data structures SYNOPSIS Without any explicit objects: use Data::Walker; Data::Walker->cli( $data_structure ); Object-style invocation: use Data::Walker; my $w = new Data::Walker; $w->walk( $data_structure ); $w->ls("-al"); $w->pwd; $w->cli; Importing methods into the current package: use Data::Walker qw(:direct); walk $data_structure; ls "-al"; pwd; cli; DESCRIPTION This module allows you to "walk" an arbitrary Perl data structure in the same way that you can walk a directory tree from a UNIX command line. It reuses familiar unix commands (such as "ls", "cd", "pwd") and applies these to data structures. It has a command-line interface which behaves like a UNIX shell. You can also use object-style sytax to invoke the CLI commands from outside the CLI. Data::Walker objects are encapsulated, so that you can hop into and out of a CLI without losing state, and you can have several Data::Walker objects pointing at different structures. The main functions can also be imported and used directly from within the Perl debugger's CLI. INSTALLATION To install this package, just into the directory which you created by untarring the package, and type the following: perl Makefile.PL make test make make install This will copy Walker.pm to your perl library directory for use by all perl scripts. You probably must be root to do this, unless you have installed a personal copy of perl or you have write access to a Perl lib directory. USAGE You open a command-line interface by invoking the cli() function. use Data::Walker; Data::Walker->cli( $data_structure ); You can customize certain features, like so: use Data::Walker; $Data::Walker::Config{'skipdoublerefs'} = 0; Data::Walker->cli( $data_structure ); If you prefer to use object-style notation, then you can use this syntax to customize the settings. You can invoke the walk() method directly, our you can let the cli() method call walk() implicitly: use Data::Walker; my $w1 = new Data::Walker; $w1->walk( $data_structure ); $w1->cli; my $w2 = new Data::Walker; $w2->cli( $data_structure ); my $w3 = new Data::Walker( 'skipdoublerefs' => 0 ); $w3->walk( $data_structure ); $w3->cli(); $w3->showrecursion(0); $w3->cli(); You can also import most of the functions directly into the current package. This is especially useful from within the debugger (see the example below). use Data::Walker qw(:direct); walk $data_structure; ls "-al"; pwd; cli; When you use the :direct pragma and invoke the walk() function, a Data::Walker object is implicitly created, and is available as $Data::Walker::WALKER. Imagine a data structure like so: my $s = { a => [ 10, 20, "thirty" ], b => { "w" => "forty", "x" => "fifty", "y" => 60, "z" => \70, }, c => sub { print "I'm a data structure!\n"; }, d => 80, }; $s->{e} = \$s->{d}; Here is a sample CLI session examining this structure ('/>' is the prompt): /> /> ls -l a ARRAY (3) b HASH (4) c CODE d scalar 80 e SCALAR 80 /> cd a /->{a}> ls -al .. HASH (5) . ARRAY (3) 0 scalar 10 1 scalar 20 2 scalar 'thirty' /->{a}> cd ../b /->{b}> ls -al .. HASH (5) . HASH (4) w scalar 'forty' x scalar 'fifty' y scalar 60 z SCALAR 70 /->{b}> cd .. /> dump b dump--> 'b' $b = { 'x' => 'fifty', 'y' => 60, 'z' => \70, 'w' => 'forty' }; /> ls -al .. HASH (5) . HASH (5) a ARRAY (3) b HASH (4) c CODE d scalar 80 e SCALAR 80 /> ! $cur->{d} += 3 eval--> $cur->{d} += 3 retv--> 83 /> ls -al .. HASH (5) . HASH (5) a ARRAY (3) b HASH (4) c CODE d scalar 83 e SCALAR 83 /> Below is a sample debugger session examining this structure. Note that the walk() function returns a reference to the "cursor", which is itself a reference to whatever is the "current directory," so to speak. The actual Data::Walker object iself is managed implicitly, and is available as $Data::Walker::WALKER. When you are finished, you can undef this object directly, or use the unwalk() function, which does this for you. But if you saved a copy of the cursor, then you will need to undef this on your own. (violet) ~/perl/walker/Data-Walker-0.18 > perl -d sample_db Loading DB routines from perl5db.pl version 1.0401 Emacs support available. Enter h or `h h' for help. main::(sample:19): d => 80, DB<1> n main::(sample:22): $s->{e} = \$s->{d}; DB<1> n main::(sample:30): 1; DB<1> use Data::Walker qw(:direct) DB<2> $cur = walk $s DB<3> pwd / DB<4> ls a b c d e DB<5> lal .. HASH (5) . HASH (5) a ARRAY (3) b HASH (4) c CODE d scalar 80 e SCALAR 80 DB<6> cd a /->{a} DB<7> ll 0 scalar 10 1 scalar 20 2 scalar 'thirty' DB<8> cd '../b' /->{b} DB<9> lal .. HASH (5) . HASH (4) w scalar 'forty' x scalar 'fifty' y scalar 60 z SCALAR 70 DB<10> cd '..' / DB<11> dump b dump--> 'b' $b = { 'x' => 'fifty', 'y' => 60, 'z' => \70, 'w' => 'forty' }; DB<12> ll a ARRAY (3) b HASH (4) c CODE d scalar 80 e SCALAR 80 DB<13> $$cur->{d} += 3 DB<14> ll a ARRAY (3) b HASH (4) c CODE d scalar 83 e SCALAR 83 DB<15> DB<16> pwd / DB<17> cli /> cd b /->{b}> ls -l w scalar 'forty' x scalar 'fifty' y scalar 60 z SCALAR 70 /->{b}> print y 60 /->{b}> print x fifty /->{b}> exit DB<18> pwd /->{b} DB<19> ll w scalar 'forty' x scalar 'fifty' y scalar 60 z SCALAR 70 DB<20> unwalk DB<21> undef $cur DB<22> The following commands are available from within the CLI. With these commands, you can navigate around the data structure as if it were a directory tree. cd like UNIX cd ls like UNIX ls (also respects options -a, -l) print prints the item as a scalar dump invokes Data::Dumper set set configuration variables show show configuration variables ! or eval eval arbitrary perl (careful!) help this help message help set lists the available config variables For each session (or object) the following items can be configured: rootname (default: '/' ) displays the root node refname (default: 'ref' ) displays embedded refs scalarname (default: 'scalar') displays simple scalars undefname (default: 'undef' ) displays undefined scalars maxdepth (default: 1 ) maximum dump-depth (Data::Dumper) indent (default: 1 ) amount of indent (Data::Dumper) lscol1width (default: 15 ) column widths for 'ls' displays lscol2width (default: 25 ) column widths for 'ls' displays showrecursion (default: 1 ) note recursion in the prompt showids (default: 0 ) show ref id numbers in ls lists skipdoublerefs (default: 1 ) hop over ref-to-refs during walks skipwarning (default: 1 ) warn when hopping over ref-to-refs truncatescalars (default: 37 ) truncate scalars in 'ls' displays autoprint (default: 1 ) print directory after chdir when not in CLI promptchar (default: '>') customize the session prompt arrowshaft (default: '-') ('-' in '->') arrowhead (default: '>') ('>' in '->') curname (default: 'cur' ) how to refer to the cursor for evals parname (default: 'par' ) how to refer to the parent ref for evals CHANGES * Version 1.05 Patch to the test scripts for compatibility with perl 5.8.0, which stringifies references-to-references differently. In previous versions of perl, references-to-references were stringified as 'SCALAR(0x???)', but perl 5.8.0 stringifies them as 'REF(0x???)'. All versions of perl's 'ref' function return 'REF' for references-to-references. * Version 1.02-1.04 Minor changes to installer tests. * Version 1.01 Minor changes to the documentation. Added walker_http.pl, which is a library for using Data::Walker together with HTTP::Daemon to view objects with a Web browser. Two example scripts are also included. * Version 0.21 Minor changes to the documentation * Version 0.19-0.20 Added new tests and updated the documentation. * Version 0.18 Completely separated the CLI loop, command-parsing regexes, and the functions which implement the commands. AUTOLOAD is now set up to handle any commands that the CLI can parse (except for eval() ). By using the :direct pragma, you can now import AUTOLOADed functions into the current package, so that you can easily invoke them from the perl debugger. * Version 0.16-0.17 The Data::Walker objects are now fully encapsulated. NOTE: The walk() function has been separated into two functions, namely walk() and cli(). The usage instructions have changed. Please have a look. * Version 0.15 Reorganized the installation tests. A few minor changes to the module itself. * Version 0.13-0.14 Moved some functionality from the CLI-loop into distinct functions. * Version 0.12 Blessed references to non-hashes are now handled correctly. Modified the output of "ls" commands (looks different). Added new options: showids, lscol2width, scalarname, undefname, skipwarning Numerous internal changes. * Version 0.11 Fixed some misspellings in the help information. Modified the pretty-print format of scalars. Added some new comments to the source code. Various other small updates. THANKS Thanks to Gurusamy Sarathy for writing Data::Dumper, and to Dominique Dumont for writing Tk::ObjScanner. Thanks to Matthew Persico for sending some ideas on how this module might be useful in the debugger. Thanks to Scott Lindsey for pointing out that this module is useful for reading files created with the Storable module, and for sending a sample script to do this. AUTHOR John Nolan jpnolan@sonic.net 1999,2000. A copyright statment is contained within the source code itself.