diff options
Diffstat (limited to 'Master/tlpkg/tlperl0/lib/Tie')
| -rwxr-xr-x | Master/tlpkg/tlperl0/lib/Tie/Array.pm | 287 | ||||
| -rwxr-xr-x | Master/tlpkg/tlperl0/lib/Tie/EncryptedHash.pm | 451 | ||||
| -rwxr-xr-x | Master/tlpkg/tlperl0/lib/Tie/File.pm | 2632 | ||||
| -rwxr-xr-x | Master/tlpkg/tlperl0/lib/Tie/Handle.pm | 201 | ||||
| -rwxr-xr-x | Master/tlpkg/tlperl0/lib/Tie/Hash.pm | 257 | ||||
| -rwxr-xr-x | Master/tlpkg/tlperl0/lib/Tie/Hash/NamedCapture.pm | 62 | ||||
| -rwxr-xr-x | Master/tlpkg/tlperl0/lib/Tie/Memoize.pm | 128 | ||||
| -rwxr-xr-x | Master/tlpkg/tlperl0/lib/Tie/RefHash.pm | 274 | ||||
| -rwxr-xr-x | Master/tlpkg/tlperl0/lib/Tie/Registry.pm | 45 | ||||
| -rwxr-xr-x | Master/tlpkg/tlperl0/lib/Tie/Scalar.pm | 140 | ||||
| -rwxr-xr-x | Master/tlpkg/tlperl0/lib/Tie/StdHandle.pm | 71 | ||||
| -rwxr-xr-x | Master/tlpkg/tlperl0/lib/Tie/SubstrHash.pm | 215 | ||||
| -rwxr-xr-x | Master/tlpkg/tlperl0/lib/Tie/Watch.pm | 560 |
13 files changed, 5323 insertions, 0 deletions
diff --git a/Master/tlpkg/tlperl0/lib/Tie/Array.pm b/Master/tlpkg/tlperl0/lib/Tie/Array.pm new file mode 100755 index 00000000000..af8f51e9f51 --- /dev/null +++ b/Master/tlpkg/tlperl0/lib/Tie/Array.pm @@ -0,0 +1,287 @@ +package Tie::Array; + +use 5.006_001; +use strict; +use Carp; +our $VERSION = '1.03'; + +# Pod documentation after __END__ below. + +sub DESTROY { } +sub EXTEND { } +sub UNSHIFT { scalar shift->SPLICE(0,0,@_) } +sub SHIFT { shift->SPLICE(0,1) } +sub CLEAR { shift->STORESIZE(0) } + +sub PUSH +{ + my $obj = shift; + my $i = $obj->FETCHSIZE; + $obj->STORE($i++, shift) while (@_); +} + +sub POP +{ + my $obj = shift; + my $newsize = $obj->FETCHSIZE - 1; + my $val; + if ($newsize >= 0) + { + $val = $obj->FETCH($newsize); + $obj->STORESIZE($newsize); + } + $val; +} + +sub SPLICE { + my $obj = shift; + my $sz = $obj->FETCHSIZE; + my $off = (@_) ? shift : 0; + $off += $sz if ($off < 0); + my $len = (@_) ? shift : $sz - $off; + $len += $sz - $off if $len < 0; + my @result; + for (my $i = 0; $i < $len; $i++) { + push(@result,$obj->FETCH($off+$i)); + } + $off = $sz if $off > $sz; + $len -= $off + $len - $sz if $off + $len > $sz; + if (@_ > $len) { + # Move items up to make room + my $d = @_ - $len; + my $e = $off+$len; + $obj->EXTEND($sz+$d); + for (my $i=$sz-1; $i >= $e; $i--) { + my $val = $obj->FETCH($i); + $obj->STORE($i+$d,$val); + } + } + elsif (@_ < $len) { + # Move items down to close the gap + my $d = $len - @_; + my $e = $off+$len; + for (my $i=$off+$len; $i < $sz; $i++) { + my $val = $obj->FETCH($i); + $obj->STORE($i-$d,$val); + } + $obj->STORESIZE($sz-$d); + } + for (my $i=0; $i < @_; $i++) { + $obj->STORE($off+$i,$_[$i]); + } + return wantarray ? @result : pop @result; +} + +sub EXISTS { + my $pkg = ref $_[0]; + croak "$pkg doesn't define an EXISTS method"; +} + +sub DELETE { + my $pkg = ref $_[0]; + croak "$pkg doesn't define a DELETE method"; +} + +package Tie::StdArray; +use vars qw(@ISA); +@ISA = 'Tie::Array'; + +sub TIEARRAY { bless [], $_[0] } +sub FETCHSIZE { scalar @{$_[0]} } +sub STORESIZE { $#{$_[0]} = $_[1]-1 } +sub STORE { $_[0]->[$_[1]] = $_[2] } +sub FETCH { $_[0]->[$_[1]] } +sub CLEAR { @{$_[0]} = () } +sub POP { pop(@{$_[0]}) } +sub PUSH { my $o = shift; push(@$o,@_) } +sub SHIFT { shift(@{$_[0]}) } +sub UNSHIFT { my $o = shift; unshift(@$o,@_) } +sub EXISTS { exists $_[0]->[$_[1]] } +sub DELETE { delete $_[0]->[$_[1]] } + +sub SPLICE +{ + my $ob = shift; + my $sz = $ob->FETCHSIZE; + my $off = @_ ? shift : 0; + $off += $sz if $off < 0; + my $len = @_ ? shift : $sz-$off; + return splice(@$ob,$off,$len,@_); +} + +1; + +__END__ + +=head1 NAME + +Tie::Array - base class for tied arrays + +=head1 SYNOPSIS + + package Tie::NewArray; + use Tie::Array; + @ISA = ('Tie::Array'); + + # mandatory methods + sub TIEARRAY { ... } + sub FETCH { ... } + sub FETCHSIZE { ... } + + sub STORE { ... } # mandatory if elements writeable + sub STORESIZE { ... } # mandatory if elements can be added/deleted + sub EXISTS { ... } # mandatory if exists() expected to work + sub DELETE { ... } # mandatory if delete() expected to work + + # optional methods - for efficiency + sub CLEAR { ... } + sub PUSH { ... } + sub POP { ... } + sub SHIFT { ... } + sub UNSHIFT { ... } + sub SPLICE { ... } + sub EXTEND { ... } + sub DESTROY { ... } + + package Tie::NewStdArray; + use Tie::Array; + + @ISA = ('Tie::StdArray'); + + # all methods provided by default + + package main; + + $object = tie @somearray,Tie::NewArray; + $object = tie @somearray,Tie::StdArray; + $object = tie @somearray,Tie::NewStdArray; + + + +=head1 DESCRIPTION + +This module provides methods for array-tying classes. See +L<perltie> for a list of the functions required in order to tie an array +to a package. The basic B<Tie::Array> package provides stub C<DESTROY>, +and C<EXTEND> methods that do nothing, stub C<DELETE> and C<EXISTS> +methods that croak() if the delete() or exists() builtins are ever called +on the tied array, and implementations of C<PUSH>, C<POP>, C<SHIFT>, +C<UNSHIFT>, C<SPLICE> and C<CLEAR> in terms of basic C<FETCH>, C<STORE>, +C<FETCHSIZE>, C<STORESIZE>. + +The B<Tie::StdArray> package provides efficient methods required for tied arrays +which are implemented as blessed references to an "inner" perl array. +It inherits from B<Tie::Array>, and should cause tied arrays to behave exactly +like standard arrays, allowing for selective overloading of methods. + +For developers wishing to write their own tied arrays, the required methods +are briefly defined below. See the L<perltie> section for more detailed +descriptive, as well as example code: + +=over 4 + +=item TIEARRAY classname, LIST + +The class method is invoked by the command C<tie @array, classname>. Associates +an array instance with the specified class. C<LIST> would represent +additional arguments (along the lines of L<AnyDBM_File> and compatriots) needed +to complete the association. The method should return an object of a class which +provides the methods below. + +=item STORE this, index, value + +Store datum I<value> into I<index> for the tied array associated with +object I<this>. If this makes the array larger then +class's mapping of C<undef> should be returned for new positions. + +=item FETCH this, index + +Retrieve the datum in I<index> for the tied array associated with +object I<this>. + +=item FETCHSIZE this + +Returns the total number of items in the tied array associated with +object I<this>. (Equivalent to C<scalar(@array)>). + +=item STORESIZE this, count + +Sets the total number of items in the tied array associated with +object I<this> to be I<count>. If this makes the array larger then +class's mapping of C<undef> should be returned for new positions. +If the array becomes smaller then entries beyond count should be +deleted. + +=item EXTEND this, count + +Informative call that array is likely to grow to have I<count> entries. +Can be used to optimize allocation. This method need do nothing. + +=item EXISTS this, key + +Verify that the element at index I<key> exists in the tied array I<this>. + +The B<Tie::Array> implementation is a stub that simply croaks. + +=item DELETE this, key + +Delete the element at index I<key> from the tied array I<this>. + +The B<Tie::Array> implementation is a stub that simply croaks. + +=item CLEAR this + +Clear (remove, delete, ...) all values from the tied array associated with +object I<this>. + +=item DESTROY this + +Normal object destructor method. + +=item PUSH this, LIST + +Append elements of LIST to the array. + +=item POP this + +Remove last element of the array and return it. + +=item SHIFT this + +Remove the first element of the array (shifting other elements down) +and return it. + +=item UNSHIFT this, LIST + +Insert LIST elements at the beginning of the array, moving existing elements +up to make room. + +=item SPLICE this, offset, length, LIST + +Perform the equivalent of C<splice> on the array. + +I<offset> is optional and defaults to zero, negative values count back +from the end of the array. + +I<length> is optional and defaults to rest of the array. + +I<LIST> may be empty. + +Returns a list of the original I<length> elements at I<offset>. + +=back + +=head1 CAVEATS + +There is no support at present for tied @ISA. There is a potential conflict +between magic entries needed to notice setting of @ISA, and those needed to +implement 'tie'. + +Very little consideration has been given to the behaviour of tied arrays +when C<$[> is not default value of zero. + +=head1 AUTHOR + +Nick Ing-Simmons E<lt>nik@tiuk.ti.comE<gt> + +=cut diff --git a/Master/tlpkg/tlperl0/lib/Tie/EncryptedHash.pm b/Master/tlpkg/tlperl0/lib/Tie/EncryptedHash.pm new file mode 100755 index 00000000000..9b9979895cf --- /dev/null +++ b/Master/tlpkg/tlperl0/lib/Tie/EncryptedHash.pm @@ -0,0 +1,451 @@ +#!/usr/bin/perl -s +## +## Tie::EncryptedHash - A tied hash with encrypted fields. +## +## Copyright (c) 2000, Vipul Ved Prakash. All rights reserved. +## This code is based on Damian Conway's Tie::SecureHash. +## +## $Id: EncryptedHash.pm,v 1.8 2000/09/02 19:23:00 vipul Exp vipul $ +## vi:expandtab=1;ts=4; + +package Tie::EncryptedHash; + +use strict; +use vars qw($VERSION $strict); +use Digest::MD5 qw(md5_base64); +use Crypt::CBC; +use Data::Dumper; +use Carp; + +( $VERSION ) = '$Revision: 1.8 $' =~ /\s(\d+\.\d+)\s/; + +my $DEBUG = 0; + +sub debug { + return undef unless $DEBUG; + my ($caller, undef) = caller; + my (undef,undef,$line,$sub) = caller(1); $sub =~ s/.*://; + $sub = sprintf "%10s()%4d",$sub,$line; + print "$sub " . (shift) . "\n"; +} + +# sub new { +# my ($class,%args) = @_; +# my %self = (); tie %self, $class; +# my $self = bless \%self, $class; +# $self->{__password} = $args{__password} if $args{__password}; +# $self->{__cipher} = $args{__cipher} || qq{Blowfish}; +# return $self; +# } + +sub new { + my ($class,%args) = @_; + my $self = {}; tie %$self, $class; + bless $self, $class; + $self->{__password} = $args{__password} if $args{__password}; + $self->{__cipher} = $args{__cipher} || qq{Blowfish}; + return $self; +} + + +sub _access { + + my ($self, $key, $caller, $file, $value, $delete) = @_; + my $class = ref $self || $self; + # SPECIAL ATTRIBUTE + if ( $key =~ /(__password|__hide|__scaffolding|__cipher)$/ ) { + my $key = $1; + unless($value||$delete) {return undef unless $caller eq $class} + if ($delete && ($key =~ /__password/)) { + for (keys %{$$self{__scaffolding}}) { + if ( ref $self->{$_} ) { + $self->{$_} = encrypt($self->{$_}, $self->{__scaffolding}{$_}, $self->{__cipher}); + delete $self->{__scaffolding}{$_}; + } + } + } + delete $$self{$key} if $delete; + return $self->{$key} = $value if $value; + return $self->{$key}; + # SECRET FIELD + } elsif ( $key =~ m/^(_{1}[^_][^:]*)$/ ||$key =~ m/.*?::(_{1}[^_][^:]*)/ ) { + my $ctext = $self->{$1}; + if ( ref $ctext && !($value)) { # ENCRYPT REF AT FETCH + my $pass = $self->{__scaffolding}{$1} || $self->{__password}; + return undef unless $pass; + $self->{$1} = encrypt($ctext, $pass, $self->{__cipher}); + return $self->FETCH ($1); + } + my $ptext = qq{}; my $isnot = !( exists $self->{$1} ); + my $auth = verify($self,$1); + return undef if !($auth) && ref $self->{$1}; + return undef if !($auth) && $self->{__hide}; + if ($auth && $auth ne "1") { $ptext = $auth } + if ($value && $auth) { # STORE + if ( ref $value ) { + $self->{__scaffolding}{$1} = $self->{__password}; $ctext = $value; + } else { + my $key = $1; + unless ($self->{__password}) { + if ($value =~ m:^\S+\s\S{22}\s:) { + return $self->{$key} = $value; + } else { return undef } + } + $ctext = encrypt($value, $self->{__password}, $self->{__cipher}); + } + $self->{$1} = $ctext; + return $value; + } elsif ($auth && $delete) { # DELETE + delete $$self{$1} + } elsif ($isnot && (!($value))) { # DOESN'T EXIST + return; + } elsif ((!($auth)) && $ctext) { + return $ctext; # FETCH return ciphertext + } elsif ($auth && !($isnot)) { # FETCH return plaintext + if (ref $ptext) { + $self->{$1} = $ptext; + $self->{__scaffolding}{$1} = $self->{__password}; # Ref counting mechanism + return $self->{$1}; + } + } + return undef unless $auth; + return $ptext; + # PUBLIC FIELD + } elsif ( $key =~ m/([^:]*)$/ || $key =~ m/.*?::([^:]*)/ ) { + $self->{$1} = $value if $value; + delete $$self{$1} if $delete; + return $self->{$1} if $self->{$1}; + return undef; + } + +} + +sub encrypt { # ($plaintext, $password, $cipher) + $_[0] = qq{REF }. Data::Dumper->new([$_[0]])->Indent(0)->Terse(0)->Purity(1)->Dumpxs if ref $_[0]; + return qq{$_[2] } . md5_base64($_[0]) .qq{ } . + Crypt::CBC->new($_[1],$_[2])->encrypt_hex($_[0]) +} + +sub decrypt { # ($cipher $md5sum $ciphertext, $password) + return undef unless $_[1]; + my ($m, $d, $c) = split /\s/,$_[0]; + my $ptext = Crypt::CBC->new($_[1],$m)->decrypt_hex($c); + my $check = md5_base64($ptext); + if ( $d eq $check ) { + if ($ptext =~ /^REF (.*)/is) { + my ($VAR1,$VAR2,$VAR3,$VAR4,$VAR5,$VAR6,$VAR7,$VAR8); + return eval qq{$1}; + } + return $ptext; + } +} + +sub verify { # ($self, $key) + my ($self, $key) = splice @_,0,2; + # debug ("$self->{__scaffolding}{$key}, $self->{__password}, $self->{$key}"); + return 1 unless $key =~ m:^_:; + return 1 unless exists $self->{$key}; + return undef if ref $self->{$key} && ($self->{__scaffolding}{$key} ne + $self->{__password}); + my $ptext = decrypt($self->{$key}, $self->{__password}); + return $ptext if $ptext; +} + +sub each { CORE::each %{$_[0]} } +sub keys { CORE::keys %{$_[0]} } +sub values { CORE::values %{$_[0]} } +sub exists { CORE::exists $_[0]->{$_[1]} } + +sub TIEHASH # ($class, @args) +{ + my $class = ref($_[0]) || $_[0]; + my $self = bless {}, $class; + $self->{__password} = $_[1] if $_[1]; + $self->{__cipher} = $_[2] || qq{Blowfish}; + return $self; +} + +sub FETCH # ($self, $key) +{ + my ($self, $key) = @_; + my $entry = _access($self,$key,(caller)[0..1]); + return $entry if $entry; +} + +sub STORE # ($self, $key, $value) +{ + my ($self, $key, $value) = @_; + my $entry = _access($self,$key,(caller)[0..1],$value); + return $entry if $entry; +} + +sub DELETE # ($self, $key) +{ + my ($self, $key) = @_; + return _access($self,$key,(caller)[0..1],'',1); +} + +sub CLEAR # ($self) +{ + my ($self) = @_; + return undef if grep { ! $self->verify($_) } + grep { ! /__/ } CORE::keys %{$self}; + %{$self} = (); +} + +sub EXISTS # ($self, $key) +{ + my ($self, $key) = @_; + my @context = (caller)[0..1]; + return _access($self,$key,@context) ? 1 : ''; +} + +sub FIRSTKEY # ($self) +{ + my ($self) = @_; + CORE::keys %{$self}; + goto &NEXTKEY; +} + +sub NEXTKEY # ($self) +{ + my $self = $_[0]; my $key; + my @context = (caller)[0..1]; + while (defined($key = CORE::each %{$self})) { + last if eval { _access($self,$key,@context) } + } + return $key; +} + +sub DESTROY # ($self) +{ +} + +1; +__END__ + + +=head1 NAME + +Tie::EncryptedHash - Hashes (and objects based on hashes) with encrypting fields. + +=head1 SYNOPSIS + + use Tie::EncryptedHash; + + my %s = (); + tie %s, Tie::EncryptedHash, 'passwd'; + + $s{foo} = "plaintext"; # Normal field, stored in plaintext. + print $s{foo}; # (plaintext) + + $s{_bar} = "signature"; # Fieldnames that begin in single + # underscore are encrypted. + print $s{_bar}; # (signature) Though, while the password + # is set, they behave like normal fields. + delete $s{__password}; # Delete password to disable access + # to encrypting fields. + print $s{_bar}; # (Blowfish NuRVFIr8UCAJu5AWY0w...) + + $s{__password} = 'passwd'; # Restore password to gain access. + print $s{_bar}; # (signature) + + $s{_baz}{a}{b} = 42; # Refs are fine, we encrypt them too. + + +=head1 DESCRIPTION + +Tie::EncryptedHash augments Perl hash semantics to build secure, encrypting +containers of data. Tie::EncryptedHash introduces special hash fields that +are coupled with encrypt/decrypt routines to encrypt assignments at STORE() +and decrypt retrievals at FETCH(). By design, I<encrypting fields> are +associated with keys that begin in single underscore. The remaining +keyspace is used for accessing normal hash fields, which are retained +without modification. + +While the password is set, a Tie::EncryptedHash behaves exactly like a +standard Perl hash. This is its I<transparent mode> of access. Encrypting +and normal fields are identical in this mode. When password is deleted, +encrypting fields are accessible only as ciphertext. This is +Tie::EncryptedHash's I<opaque mode> of access, optimized for serialization. + +Encryption is done with Crypt::CBC(3) which encrypts in the cipher block +chaining mode with Blowfish, DES or IDEA. Tie::EncryptedHash uses Blowfish +by default, but can be instructed to employ any cipher supported by +Crypt::CBC(3). + +=head1 MOTIVATION + +Tie::EncryptedHash was designed for storage and communication of key +material used in public key cryptography algorithms. I abstracted out the +mechanism for encrypting selected fields of a structured data record because +of the sheer convenience of this data security method. + +Quite often, applications that require data confidentiality eschew strong +cryptography in favor of OS-based access control mechanisms because of the +additional costs of cryptography integration. Besides cipher +implementations, which are available as ready-to-deploy perl modules, use of +cryptography in an application requires code to aid conversion and +representation of encrypted data. This code is usually encapsulated in a +data access layer that manages encryption, decryption, access control and +re-structuring of flat plaintext according to a data model. +Tie::EncryptedHash provides these functions under the disguise of a Perl +hash so perl applications can use strong cryptography without the cost of +implementing a complex data access layer. + + +=head1 CONSTRUCTION + +=head2 Tied Hash + +C<tie %h, Tie::EncryptedHash, 'Password', 'Cipher';> + +Ties %h to Tie::EncryptedHash and sets the value of password and cipher to +'Password' and 'Cipher'. Both arguments are optional. + +=head2 Blessed Object + +C<$h = new Tie::EncryptedHash __password => 'Password', + __cipher => 'Cipher';> + +The new() constructor returns an object that is both tied and blessed into +Tie::EncryptedHash. Both arguments are optional. When used in this manner, +Tie::EncryptedHash behaves like a class with encrypting data members. + +=head1 RESERVED ATTRIBUTES + +The attributes __password, __cipher and __hide are reserved for +communication with object methods. They are "write-only" from everywhere +except the class to which the hash is tied. __scaffolding is inaccessible. +Tie::EncryptedHash stores the current encryption password and some transient +data structures in these fields and restricts access to them on need-to-know +basis. + +=head2 __password + +C<$h{__password} = "new password"; +delete $h{__password};> + +The password is stored under the attribute C<__password>. In addition to +specifying a password at construction, assigning to the __password attribute +sets the current encryption password to the assigned value. Deleting the +__password unsets it and switches the hash into opaque mode. + +=head2 __cipher + +C<$h{__cipher} = 'DES'; $h{__cipher} = 'Blowfish';> + +The cipher used for encryption/decryption is stored under the attribute +__cipher. The value defaults to 'Blowfish'. + +=head2 __hide + +C<$h{__hide} = 1;> + +Setting this attribute I<hides> encrypting fields in opaque mode. 'undef' +is returned at FETCH() and EXISTS(). + +=head1 BEHAVIOR + +=head2 References + +A reference stored in an encrypting field is serialized before encryption. +The data structure represented by the reference is folded into a single line +of ciphertext which is stored under the first level key. In the opaque +mode, therefore, only the first level of keys of the hash will be visible. + +=head2 Opaque Mode + +The opaque mode introduces several other constraints on access of encrypting +fields. Encrypting fields return ciphertext on FETCH() unless __hide +attribute is set, which forces Tie::EncryptedHash to behave as if encrypting +fields don't exist. Irrespective of __hide, however, DELETE() and CLEAR() +fail in opaque mode. So does STORE() on an existing encrypting field. +Plaintext assignments to encrypting fields are silently ignored, but +ciphertext assignments are fine. Ciphertext assignments can be used to move +data between different EncryptedHashes. + +=head2 Multiple Passwords and Ciphers + +Modality of Tie::EncryptedHash's access system breaks down when more than +one password is used to with different encrypting fields. This is a +feature. Tie::EncryptedHash lets you mix passwords and ciphers in the same +hash. Assign new values to __password and __cipher and create a new +encrypting field. Transparent mode will be restricted to fields encrypted +with the current password. + +=head2 Error Handling + +Tie::Encrypted silently ignores access errors. It doesn't carp/croak when +you perform an illegal operation (like assign plaintext to an encrypting +field in opaque mode). This is to prevent data lossage, the kind that +results from abnormal termination of applications. + +=head1 QUIRKS + +=head2 Autovivification + +Due to the nature of autovivified references (which spring into existence +when an undefined reference is dereferenced), references are stored as +plaintext in transparent mode. Analogous ciphertext representations are +maintained in parallel and restored to encrypting fields when password is +deleted. This process is completely transparent to the user, though it's +advisable to delete the password after the final assignment to a +Tie::EncryptedHash. This ensures plaintext representations and scaffolding +data structures are duly flushed. + +=head2 Data::Dumper + +Serialization of references is done with Data::Dumper, therefore the nature +of data that can be assigned to encrypting fields is limited by what +Data::Dumper can grok. We set $Data::Dumper::Purity = 1, so +self-referential and recursive structures should be OK. + +=head2 Speed + +Tie::EncryptedHash'es keep their contents encrypted as much as possible, so +there's a rather severe speed penalty. With Blowfish, STORE() on +EncryptedHash can be upto 70 times slower than a standard perl hash. +Reference STORE()'es will be quicker, but speed gain will be adjusted at +FETCH(). FETCH() is about 35 times slower than a standard perl hash. DES +affords speed improvements of upto 2x, but is not considered secure for +long-term storage of data. These values were computed on a DELL PIII-300 +Mhz notebook with 128 Mb RAM running perl 5.003 on Linux 2.2.16. Variations +in speed might be different on your machine. + +=head1 STANDARD USAGE + +The standard usage for this module would be something along the lines of: +populate Tie::EncryptedHash with sensitive data, delete the password, +serialize the encrypted hash with Data::Dumper, store the result on disk or +send it over the wire to another machine. Later, when the sensitive data is +required, procure the EncryptedHash, set the password and accesses the +encrypted data fields. + +=head1 SEE ALSO + +Data::Dumper(3), +Crypt::CBC(3), +Crypt::DES(3), +Crypt::Blowfish(3), +Tie::SecureHash(3) + +=head1 ACKNOWLEDGEMENTS + +The framework of Tie::EncryptedHash derives heavily from Damian Conway's +Tie::SecureHash. Objects that are blessed as well as tied are just one of +the pleasant side-effects of stealing Damian's code. Thanks to Damian for +this brilliant module. + +PacificNet (http://www.pacificnet.net) loaned me the aforementioned notebook +to hack from the comfort of my bed. Thanks folks! + +=head1 AUTHOR + +Vipul Ved Prakash <mail@vipul.net> + +=head1 LICENSE + +This module is distributed under the same license as Perl itself. + + diff --git a/Master/tlpkg/tlperl0/lib/Tie/File.pm b/Master/tlpkg/tlperl0/lib/Tie/File.pm new file mode 100755 index 00000000000..9528ab1bfdb --- /dev/null +++ b/Master/tlpkg/tlperl0/lib/Tie/File.pm @@ -0,0 +1,2632 @@ + +package Tie::File; +require 5.005; +use Carp ':DEFAULT', 'confess'; +use POSIX 'SEEK_SET'; +use Fcntl 'O_CREAT', 'O_RDWR', 'LOCK_EX', 'LOCK_SH', 'O_WRONLY', 'O_RDONLY'; +sub O_ACCMODE () { O_RDONLY | O_RDWR | O_WRONLY } + + +$VERSION = "0.97_02"; +my $DEFAULT_MEMORY_SIZE = 1<<21; # 2 megabytes +my $DEFAULT_AUTODEFER_THRESHHOLD = 3; # 3 records +my $DEFAULT_AUTODEFER_FILELEN_THRESHHOLD = 65536; # 16 disk blocksful + +my %good_opt = map {$_ => 1, "-$_" => 1} + qw(memory dw_size mode recsep discipline + autodefer autochomp autodefer_threshhold concurrent); + +sub TIEARRAY { + if (@_ % 2 != 0) { + croak "usage: tie \@array, $_[0], filename, [option => value]..."; + } + my ($pack, $file, %opts) = @_; + + # transform '-foo' keys into 'foo' keys + for my $key (keys %opts) { + unless ($good_opt{$key}) { + croak("$pack: Unrecognized option '$key'\n"); + } + my $okey = $key; + if ($key =~ s/^-+//) { + $opts{$key} = delete $opts{$okey}; + } + } + + if ($opts{concurrent}) { + croak("$pack: concurrent access not supported yet\n"); + } + + unless (defined $opts{memory}) { + # default is the larger of the default cache size and the + # deferred-write buffer size (if specified) + $opts{memory} = $DEFAULT_MEMORY_SIZE; + $opts{memory} = $opts{dw_size} + if defined $opts{dw_size} && $opts{dw_size} > $DEFAULT_MEMORY_SIZE; + # Dora Winifred Read + } + $opts{dw_size} = $opts{memory} unless defined $opts{dw_size}; + if ($opts{dw_size} > $opts{memory}) { + croak("$pack: dw_size may not be larger than total memory allocation\n"); + } + # are we in deferred-write mode? + $opts{defer} = 0 unless defined $opts{defer}; + $opts{deferred} = {}; # no records are presently deferred + $opts{deferred_s} = 0; # count of total bytes in ->{deferred} + $opts{deferred_max} = -1; # empty + + # What's a good way to arrange that this class can be overridden? + $opts{cache} = Tie::File::Cache->new($opts{memory}); + + # autodeferment is enabled by default + $opts{autodefer} = 1 unless defined $opts{autodefer}; + $opts{autodeferring} = 0; # but is not initially active + $opts{ad_history} = []; + $opts{autodefer_threshhold} = $DEFAULT_AUTODEFER_THRESHHOLD + unless defined $opts{autodefer_threshhold}; + $opts{autodefer_filelen_threshhold} = $DEFAULT_AUTODEFER_FILELEN_THRESHHOLD + unless defined $opts{autodefer_filelen_threshhold}; + + $opts{offsets} = [0]; + $opts{filename} = $file; + unless (defined $opts{recsep}) { + $opts{recsep} = _default_recsep(); + } + $opts{recseplen} = length($opts{recsep}); + if ($opts{recseplen} == 0) { + croak "Empty record separator not supported by $pack"; + } + + $opts{autochomp} = 1 unless defined $opts{autochomp}; + + $opts{mode} = O_CREAT|O_RDWR unless defined $opts{mode}; + $opts{rdonly} = (($opts{mode} & O_ACCMODE) == O_RDONLY); + $opts{sawlastrec} = undef; + + my $fh; + + if (UNIVERSAL::isa($file, 'GLOB')) { + # We use 1 here on the theory that some systems + # may not indicate failure if we use 0. + # MSWin32 does not indicate failure with 0, but I don't know if + # it will indicate failure with 1 or not. + unless (seek $file, 1, SEEK_SET) { + croak "$pack: your filehandle does not appear to be seekable"; + } + seek $file, 0, SEEK_SET; # put it back + $fh = $file; # setting binmode is the user's problem + } elsif (ref $file) { + croak "usage: tie \@array, $pack, filename, [option => value]..."; + } else { + # $fh = \do { local *FH }; # XXX this is buggy + if ($] < 5.006) { + # perl 5.005 and earlier don't autovivify filehandles + require Symbol; + $fh = Symbol::gensym(); + } + sysopen $fh, $file, $opts{mode}, 0666 or return; + binmode $fh; + ++$opts{ourfh}; + } + { my $ofh = select $fh; $| = 1; select $ofh } # autoflush on write + if (defined $opts{discipline} && $] >= 5.006) { + # This avoids a compile-time warning under 5.005 + eval 'binmode($fh, $opts{discipline})'; + croak $@ if $@ =~ /unknown discipline/i; + die if $@; + } + $opts{fh} = $fh; + + bless \%opts => $pack; +} + +sub FETCH { + my ($self, $n) = @_; + my $rec; + + # check the defer buffer + $rec = $self->{deferred}{$n} if exists $self->{deferred}{$n}; + $rec = $self->_fetch($n) unless defined $rec; + + # inlined _chomp1 + substr($rec, - $self->{recseplen}) = "" + if defined $rec && $self->{autochomp}; + $rec; +} + +# Chomp many records in-place; return nothing useful +sub _chomp { + my $self = shift; + return unless $self->{autochomp}; + if ($self->{autochomp}) { + for (@_) { + next unless defined; + substr($_, - $self->{recseplen}) = ""; + } + } +} + +# Chomp one record in-place; return modified record +sub _chomp1 { + my ($self, $rec) = @_; + return $rec unless $self->{autochomp}; + return unless defined $rec; + substr($rec, - $self->{recseplen}) = ""; + $rec; +} + +sub _fetch { + my ($self, $n) = @_; + + # check the record cache + { my $cached = $self->{cache}->lookup($n); + return $cached if defined $cached; + } + + if ($#{$self->{offsets}} < $n) { + return if $self->{eof}; # request for record beyond end of file + my $o = $self->_fill_offsets_to($n); + # If it's still undefined, there is no such record, so return 'undef' + return unless defined $o; + } + + my $fh = $self->{FH}; + $self->_seek($n); # we can do this now that offsets is populated + my $rec = $self->_read_record; + +# If we happen to have just read the first record, check to see if +# the length of the record matches what 'tell' says. If not, Tie::File +# won't work, and should drop dead. +# +# if ($n == 0 && defined($rec) && tell($self->{fh}) != length($rec)) { +# if (defined $self->{discipline}) { +# croak "I/O discipline $self->{discipline} not supported"; +# } else { +# croak "File encoding not supported"; +# } +# } + + $self->{cache}->insert($n, $rec) if defined $rec && not $self->{flushing}; + $rec; +} + +sub STORE { + my ($self, $n, $rec) = @_; + die "STORE called from _check_integrity!" if $DIAGNOSTIC; + + $self->_fixrecs($rec); + + if ($self->{autodefer}) { + $self->_annotate_ad_history($n); + } + + return $self->_store_deferred($n, $rec) if $self->_is_deferring; + + + # We need this to decide whether the new record will fit + # It incidentally populates the offsets table + # Note we have to do this before we alter the cache + # 20020324 Wait, but this DOES alter the cache. TODO BUG? + my $oldrec = $self->_fetch($n); + + if (not defined $oldrec) { + # We're storing a record beyond the end of the file + $self->_extend_file_to($n+1); + $oldrec = $self->{recsep}; + } +# return if $oldrec eq $rec; # don't bother + my $len_diff = length($rec) - length($oldrec); + + # length($oldrec) here is not consistent with text mode TODO XXX BUG + $self->_mtwrite($rec, $self->{offsets}[$n], length($oldrec)); + $self->_oadjust([$n, 1, $rec]); + $self->{cache}->update($n, $rec); +} + +sub _store_deferred { + my ($self, $n, $rec) = @_; + $self->{cache}->remove($n); + my $old_deferred = $self->{deferred}{$n}; + + if (defined $self->{deferred_max} && $n > $self->{deferred_max}) { + $self->{deferred_max} = $n; + } + $self->{deferred}{$n} = $rec; + + my $len_diff = length($rec); + $len_diff -= length($old_deferred) if defined $old_deferred; + $self->{deferred_s} += $len_diff; + $self->{cache}->adj_limit(-$len_diff); + if ($self->{deferred_s} > $self->{dw_size}) { + $self->_flush; + } elsif ($self->_cache_too_full) { + $self->_cache_flush; + } +} + +# Remove a single record from the deferred-write buffer without writing it +# The record need not be present +sub _delete_deferred { + my ($self, $n) = @_; + my $rec = delete $self->{deferred}{$n}; + return unless defined $rec; + + if (defined $self->{deferred_max} + && $n == $self->{deferred_max}) { + undef $self->{deferred_max}; + } + + $self->{deferred_s} -= length $rec; + $self->{cache}->adj_limit(length $rec); +} + +sub FETCHSIZE { + my $self = shift; + my $n = $self->{eof} ? $#{$self->{offsets}} : $self->_fill_offsets; + + my $top_deferred = $self->_defer_max; + $n = $top_deferred+1 if defined $top_deferred && $n < $top_deferred+1; + $n; +} + +sub STORESIZE { + my ($self, $len) = @_; + + if ($self->{autodefer}) { + $self->_annotate_ad_history('STORESIZE'); + } + + my $olen = $self->FETCHSIZE; + return if $len == $olen; # Woo-hoo! + + # file gets longer + if ($len > $olen) { + if ($self->_is_deferring) { + for ($olen .. $len-1) { + $self->_store_deferred($_, $self->{recsep}); + } + } else { + $self->_extend_file_to($len); + } + return; + } + + # file gets shorter + if ($self->_is_deferring) { + # TODO maybe replace this with map-plus-assignment? + for (grep $_ >= $len, keys %{$self->{deferred}}) { + $self->_delete_deferred($_); + } + $self->{deferred_max} = $len-1; + } + + $self->_seek($len); + $self->_chop_file; + $#{$self->{offsets}} = $len; +# $self->{offsets}[0] = 0; # in case we just chopped this + + $self->{cache}->remove(grep $_ >= $len, $self->{cache}->ckeys); +} + +### OPTIMIZE ME +### It should not be necessary to do FETCHSIZE +### Just seek to the end of the file. +sub PUSH { + my $self = shift; + $self->SPLICE($self->FETCHSIZE, scalar(@_), @_); + + # No need to return: + # $self->FETCHSIZE; # because av.c takes care of this for me +} + +sub POP { + my $self = shift; + my $size = $self->FETCHSIZE; + return if $size == 0; +# print STDERR "# POPPITY POP POP POP\n"; + scalar $self->SPLICE($size-1, 1); +} + +sub SHIFT { + my $self = shift; + scalar $self->SPLICE(0, 1); +} + +sub UNSHIFT { + my $self = shift; + $self->SPLICE(0, 0, @_); + # $self->FETCHSIZE; # av.c takes care of this for me +} + +sub CLEAR { + my $self = shift; + + if ($self->{autodefer}) { + $self->_annotate_ad_history('CLEAR'); + } + + $self->_seekb(0); + $self->_chop_file; + $self->{cache}->set_limit($self->{memory}); + $self->{cache}->empty; + @{$self->{offsets}} = (0); + %{$self->{deferred}}= (); + $self->{deferred_s} = 0; + $self->{deferred_max} = -1; +} + +sub EXTEND { + my ($self, $n) = @_; + + # No need to pre-extend anything in this case + return if $self->_is_deferring; + + $self->_fill_offsets_to($n); + $self->_extend_file_to($n); +} + +sub DELETE { + my ($self, $n) = @_; + + if ($self->{autodefer}) { + $self->_annotate_ad_history('DELETE'); + } + + my $lastrec = $self->FETCHSIZE-1; + my $rec = $self->FETCH($n); + $self->_delete_deferred($n) if $self->_is_deferring; + if ($n == $lastrec) { + $self->_seek($n); + $self->_chop_file; + $#{$self->{offsets}}--; + $self->{cache}->remove($n); + # perhaps in this case I should also remove trailing null records? + # 20020316 + # Note that delete @a[-3..-1] deletes the records in the wrong order, + # so we only chop the very last one out of the file. We could repair this + # by tracking deleted records inside the object. + } elsif ($n < $lastrec) { + $self->STORE($n, ""); + } + $rec; +} + +sub EXISTS { + my ($self, $n) = @_; + return 1 if exists $self->{deferred}{$n}; + $n < $self->FETCHSIZE; +} + +sub SPLICE { + my $self = shift; + + if ($self->{autodefer}) { + $self->_annotate_ad_history('SPLICE'); + } + + $self->_flush if $self->_is_deferring; # move this up? + if (wantarray) { + $self->_chomp(my @a = $self->_splice(@_)); + @a; + } else { + $self->_chomp1(scalar $self->_splice(@_)); + } +} + +sub DESTROY { + my $self = shift; + $self->flush if $self->_is_deferring; + $self->{cache}->delink if defined $self->{cache}; # break circular link + if ($self->{fh} and $self->{ourfh}) { + delete $self->{ourfh}; + close delete $self->{fh}; + } +} + +sub _splice { + my ($self, $pos, $nrecs, @data) = @_; + my @result; + + $pos = 0 unless defined $pos; + + # Deal with negative and other out-of-range positions + # Also set default for $nrecs + { + my $oldsize = $self->FETCHSIZE; + $nrecs = $oldsize unless defined $nrecs; + my $oldpos = $pos; + + if ($pos < 0) { + $pos += $oldsize; + if ($pos < 0) { + croak "Modification of non-creatable array value attempted, subscript $oldpos"; + } + } + + if ($pos > $oldsize) { + return unless @data; + $pos = $oldsize; # This is what perl does for normal arrays + } + + # The manual is very unclear here + if ($nrecs < 0) { + $nrecs = $oldsize - $pos + $nrecs; + $nrecs = 0 if $nrecs < 0; + } + + # nrecs is too big---it really means "until the end" + # 20030507 + if ($nrecs + $pos > $oldsize) { + $nrecs = $oldsize - $pos; + } + } + + $self->_fixrecs(@data); + my $data = join '', @data; + my $datalen = length $data; + my $oldlen = 0; + + # compute length of data being removed + for ($pos .. $pos+$nrecs-1) { + last unless defined $self->_fill_offsets_to($_); + my $rec = $self->_fetch($_); + last unless defined $rec; + push @result, $rec; + + # Why don't we just use length($rec) here? + # Because that record might have come from the cache. _splice + # might have been called to flush out the deferred-write records, + # and in this case length($rec) is the length of the record to be + # *written*, not the length of the actual record in the file. But + # the offsets are still true. 20020322 + $oldlen += $self->{offsets}[$_+1] - $self->{offsets}[$_] + if defined $self->{offsets}[$_+1]; + } + $self->_fill_offsets_to($pos+$nrecs); + + # Modify the file + $self->_mtwrite($data, $self->{offsets}[$pos], $oldlen); + # Adjust the offsets table + $self->_oadjust([$pos, $nrecs, @data]); + + { # Take this read cache stuff out into a separate function + # You made a half-attempt to put it into _oadjust. + # Finish something like that up eventually. + # STORE also needs to do something similarish + + # update the read cache, part 1 + # modified records + for ($pos .. $pos+$nrecs-1) { + my $new = $data[$_-$pos]; + if (defined $new) { + $self->{cache}->update($_, $new); + } else { + $self->{cache}->remove($_); + } + } + + # update the read cache, part 2 + # moved records - records past the site of the change + # need to be renumbered + # Maybe merge this with the previous block? + { + my @oldkeys = grep $_ >= $pos + $nrecs, $self->{cache}->ckeys; + my @newkeys = map $_-$nrecs+@data, @oldkeys; + $self->{cache}->rekey(\@oldkeys, \@newkeys); + } + + # Now there might be too much data in the cache, if we spliced out + # some short records and spliced in some long ones. If so, flush + # the cache. + $self->_cache_flush; + } + + # Yes, the return value of 'splice' *is* actually this complicated + wantarray ? @result : @result ? $result[-1] : undef; +} + + +# write data into the file +# $data is the data to be written. +# it should be written at position $pos, and should overwrite +# exactly $len of the following bytes. +# Note that if length($data) > $len, the subsequent bytes will have to +# be moved up, and if length($data) < $len, they will have to +# be moved down +sub _twrite { + my ($self, $data, $pos, $len) = @_; + + unless (defined $pos) { + die "\$pos was undefined in _twrite"; + } + + my $len_diff = length($data) - $len; + + if ($len_diff == 0) { # Woo-hoo! + my $fh = $self->{fh}; + $self->_seekb($pos); + $self->_write_record($data); + return; # well, that was easy. + } + + # the two records are of different lengths + # our strategy here: rewrite the tail of the file, + # reading ahead one buffer at a time + # $bufsize is required to be at least as large as the data we're overwriting + my $bufsize = _bufsize($len_diff); + my ($writepos, $readpos) = ($pos, $pos+$len); + my $next_block; + my $more_data; + + # Seems like there ought to be a way to avoid the repeated code + # and the special case here. The read(1) is also a little weird. + # Think about this. + do { + $self->_seekb($readpos); + my $br = read $self->{fh}, $next_block, $bufsize; + $more_data = read $self->{fh}, my($dummy), 1; + $self->_seekb($writepos); + $self->_write_record($data); + $readpos += $br; + $writepos += length $data; + $data = $next_block; + } while $more_data; + $self->_seekb($writepos); + $self->_write_record($next_block); + + # There might be leftover data at the end of the file + $self->_chop_file if $len_diff < 0; +} + +# _iwrite(D, S, E) +# Insert text D at position S. +# Let C = E-S-|D|. If C < 0; die. +# Data in [S,S+C) is copied to [S+D,S+D+C) = [S+D,E). +# Data in [S+C = E-D, E) is returned. Data in [E, oo) is untouched. +# +# In a later version, don't read the entire intervening area into +# memory at once; do the copying block by block. +sub _iwrite { + my $self = shift; + my ($D, $s, $e) = @_; + my $d = length $D; + my $c = $e-$s-$d; + local *FH = $self->{fh}; + confess "Not enough space to insert $d bytes between $s and $e" + if $c < 0; + confess "[$s,$e) is an invalid insertion range" if $e < $s; + + $self->_seekb($s); + read FH, my $buf, $e-$s; + + $D .= substr($buf, 0, $c, ""); + + $self->_seekb($s); + $self->_write_record($D); + + return $buf; +} + +# Like _twrite, but the data-pos-len triple may be repeated; you may +# write several chunks. All the writing will be done in +# one pass. Chunks SHALL be in ascending order and SHALL NOT overlap. +sub _mtwrite { + my $self = shift; + my $unwritten = ""; + my $delta = 0; + + @_ % 3 == 0 + or die "Arguments to _mtwrite did not come in groups of three"; + + while (@_) { + my ($data, $pos, $len) = splice @_, 0, 3; + my $end = $pos + $len; # The OLD end of the segment to be replaced + $data = $unwritten . $data; + $delta -= length($unwritten); + $unwritten = ""; + $pos += $delta; # This is where the data goes now + my $dlen = length $data; + $self->_seekb($pos); + if ($len >= $dlen) { # the data will fit + $self->_write_record($data); + $delta += ($dlen - $len); # everything following moves down by this much + $data = ""; # All the data in the buffer has been written + } else { # won't fit + my $writable = substr($data, 0, $len - $delta, ""); + $self->_write_record($writable); + $delta += ($dlen - $len); # everything following moves down by this much + } + + # At this point we've written some but maybe not all of the data. + # There might be a gap to close up, or $data might still contain a + # bunch of unwritten data that didn't fit. + my $ndlen = length $data; + if ($delta == 0) { + $self->_write_record($data); + } elsif ($delta < 0) { + # upcopy (close up gap) + if (@_) { + $self->_upcopy($end, $end + $delta, $_[1] - $end); + } else { + $self->_upcopy($end, $end + $delta); + } + } else { + # downcopy (insert data that didn't fit; replace this data in memory + # with _later_ data that doesn't fit) + if (@_) { + $unwritten = $self->_downcopy($data, $end, $_[1] - $end); + } else { + # Make the file longer to accommodate the last segment that doesn' + $unwritten = $self->_downcopy($data, $end); + } + } + } +} + +# Copy block of data of length $len from position $spos to position $dpos +# $dpos must be <= $spos +# +# If $len is undefined, go all the way to the end of the file +# and then truncate it ($spos - $dpos bytes will be removed) +sub _upcopy { + my $blocksize = 8192; + my ($self, $spos, $dpos, $len) = @_; + if ($dpos > $spos) { + die "source ($spos) was upstream of destination ($dpos) in _upcopy"; + } elsif ($dpos == $spos) { + return; + } + + while (! defined ($len) || $len > 0) { + my $readsize = ! defined($len) ? $blocksize + : $len > $blocksize ? $blocksize + : $len; + + my $fh = $self->{fh}; + $self->_seekb($spos); + my $bytes_read = read $fh, my($data), $readsize; + $self->_seekb($dpos); + if ($data eq "") { + $self->_chop_file; + last; + } + $self->_write_record($data); + $spos += $bytes_read; + $dpos += $bytes_read; + $len -= $bytes_read if defined $len; + } +} + +# Write $data into a block of length $len at position $pos, +# moving everything in the block forwards to make room. +# Instead of writing the last length($data) bytes from the block +# (because there isn't room for them any longer) return them. +# +# Undefined $len means 'until the end of the file' +sub _downcopy { + my $blocksize = 8192; + my ($self, $data, $pos, $len) = @_; + my $fh = $self->{fh}; + + while (! defined $len || $len > 0) { + my $readsize = ! defined($len) ? $blocksize + : $len > $blocksize? $blocksize : $len; + $self->_seekb($pos); + read $fh, my($old), $readsize; + my $last_read_was_short = length($old) < $readsize; + $data .= $old; + my $writable; + if ($last_read_was_short) { + # If last read was short, then $data now contains the entire rest + # of the file, so there's no need to write only one block of it + $writable = $data; + $data = ""; + } else { + $writable = substr($data, 0, $readsize, ""); + } + last if $writable eq ""; + $self->_seekb($pos); + $self->_write_record($writable); + last if $last_read_was_short && $data eq ""; + $len -= $readsize if defined $len; + $pos += $readsize; + } + return $data; +} + +# Adjust the object data structures following an '_mtwrite' +# Arguments are +# [$pos, $nrecs, @length] items +# indicating that $nrecs records were removed at $recpos (a record offset) +# and replaced with records of length @length... +# Arguments guarantee that $recpos is strictly increasing. +# No return value +sub _oadjust { + my $self = shift; + my $delta = 0; + my $delta_recs = 0; + my $prev_end = -1; + my %newkeys; + + for (@_) { + my ($pos, $nrecs, @data) = @$_; + $pos += $delta_recs; + + # Adjust the offsets of the records after the previous batch up + # to the first new one of this batch + for my $i ($prev_end+2 .. $pos - 1) { + $self->{offsets}[$i] += $delta; + $newkey{$i} = $i + $delta_recs; + } + + $prev_end = $pos + @data - 1; # last record moved on this pass + + # Remove the offsets for the removed records; + # replace with the offsets for the inserted records + my @newoff = ($self->{offsets}[$pos] + $delta); + for my $i (0 .. $#data) { + my $newlen = length $data[$i]; + push @newoff, $newoff[$i] + $newlen; + $delta += $newlen; + } + + for my $i ($pos .. $pos+$nrecs-1) { + last if $i+1 > $#{$self->{offsets}}; + my $oldlen = $self->{offsets}[$i+1] - $self->{offsets}[$i]; + $delta -= $oldlen; + } + +# # also this data has changed, so update it in the cache +# for (0 .. $#data) { +# $self->{cache}->update($pos + $_, $data[$_]); +# } +# if ($delta_recs) { +# my @oldkeys = grep $_ >= $pos + @data, $self->{cache}->ckeys; +# my @newkeys = map $_ + $delta_recs, @oldkeys; +# $self->{cache}->rekey(\@oldkeys, \@newkeys); +# } + + # replace old offsets with new + splice @{$self->{offsets}}, $pos, $nrecs+1, @newoff; + # What if we just spliced out the end of the offsets table? + # shouldn't we clear $self->{eof}? Test for this XXX BUG TODO + + $delta_recs += @data - $nrecs; # net change in total number of records + } + + # The trailing records at the very end of the file + if ($delta) { + for my $i ($prev_end+2 .. $#{$self->{offsets}}) { + $self->{offsets}[$i] += $delta; + } + } + + # If we scrubbed out all known offsets, regenerate the trivial table + # that knows that the file does indeed start at 0. + $self->{offsets}[0] = 0 unless @{$self->{offsets}}; + # If the file got longer, the offsets table is no longer complete + # $self->{eof} = 0 if $delta_recs > 0; + + # Now there might be too much data in the cache, if we spliced out + # some short records and spliced in some long ones. If so, flush + # the cache. + $self->_cache_flush; +} + +# If a record does not already end with the appropriate terminator +# string, append one. +sub _fixrecs { + my $self = shift; + for (@_) { + $_ = "" unless defined $_; + $_ .= $self->{recsep} + unless substr($_, - $self->{recseplen}) eq $self->{recsep}; + } +} + + +################################################################ +# +# Basic read, write, and seek +# + +# seek to the beginning of record #$n +# Assumes that the offsets table is already correctly populated +# +# Note that $n=-1 has a special meaning here: It means the start of +# the last known record; this may or may not be the very last record +# in the file, depending on whether the offsets table is fully populated. +# +sub _seek { + my ($self, $n) = @_; + my $o = $self->{offsets}[$n]; + defined($o) + or confess("logic error: undefined offset for record $n"); + seek $self->{fh}, $o, SEEK_SET + or confess "Couldn't seek filehandle: $!"; # "Should never happen." +} + +# seek to byte $b in the file +sub _seekb { + my ($self, $b) = @_; + seek $self->{fh}, $b, SEEK_SET + or die "Couldn't seek filehandle: $!"; # "Should never happen." +} + +# populate the offsets table up to the beginning of record $n +# return the offset of record $n +sub _fill_offsets_to { + my ($self, $n) = @_; + + return $self->{offsets}[$n] if $self->{eof}; + + my $fh = $self->{fh}; + local *OFF = $self->{offsets}; + my $rec; + + until ($#OFF >= $n) { + $self->_seek(-1); # tricky -- see comment at _seek + $rec = $self->_read_record; + if (defined $rec) { + push @OFF, int(tell $fh); # Tels says that int() saves memory here + } else { + $self->{eof} = 1; + return; # It turns out there is no such record + } + } + + # we have now read all the records up to record n-1, + # so we can return the offset of record n + $OFF[$n]; +} + +sub _fill_offsets { + my ($self) = @_; + + my $fh = $self->{fh}; + local *OFF = $self->{offsets}; + + $self->_seek(-1); # tricky -- see comment at _seek + + # Tels says that inlining read_record() would make this loop + # five times faster. 20030508 + while ( defined $self->_read_record()) { + # int() saves us memory here + push @OFF, int(tell $fh); + } + + $self->{eof} = 1; + $#OFF; +} + +# assumes that $rec is already suitably terminated +sub _write_record { + my ($self, $rec) = @_; + my $fh = $self->{fh}; + local $\ = ""; + print $fh $rec + or die "Couldn't write record: $!"; # "Should never happen." +# $self->{_written} += length($rec); +} + +sub _read_record { + my $self = shift; + my $rec; + { local $/ = $self->{recsep}; + my $fh = $self->{fh}; + $rec = <$fh>; + } + return unless defined $rec; + if (substr($rec, -$self->{recseplen}) ne $self->{recsep}) { + # improperly terminated final record --- quietly fix it. +# my $ac = substr($rec, -$self->{recseplen}); +# $ac =~ s/\n/\\n/g; + $self->{sawlastrec} = 1; + unless ($self->{rdonly}) { + local $\ = ""; + my $fh = $self->{fh}; + print $fh $self->{recsep}; + } + $rec .= $self->{recsep}; + } +# $self->{_read} += length($rec) if defined $rec; + $rec; +} + +sub _rw_stats { + my $self = shift; + @{$self}{'_read', '_written'}; +} + +################################################################ +# +# Read cache management + +sub _cache_flush { + my ($self) = @_; + $self->{cache}->reduce_size_to($self->{memory} - $self->{deferred_s}); +} + +sub _cache_too_full { + my $self = shift; + $self->{cache}->bytes + $self->{deferred_s} >= $self->{memory}; +} + +################################################################ +# +# File custodial services +# + + +# We have read to the end of the file and have the offsets table +# entirely populated. Now we need to write a new record beyond +# the end of the file. We prepare for this by writing +# empty records into the file up to the position we want +# +# assumes that the offsets table already contains the offset of record $n, +# if it exists, and extends to the end of the file if not. +sub _extend_file_to { + my ($self, $n) = @_; + $self->_seek(-1); # position after the end of the last record + my $pos = $self->{offsets}[-1]; + + # the offsets table has one entry more than the total number of records + my $extras = $n - $#{$self->{offsets}}; + + # Todo : just use $self->{recsep} x $extras here? + while ($extras-- > 0) { + $self->_write_record($self->{recsep}); + push @{$self->{offsets}}, int(tell $self->{fh}); + } +} + +# Truncate the file at the current position +sub _chop_file { + my $self = shift; + truncate $self->{fh}, tell($self->{fh}); +} + + +# compute the size of a buffer suitable for moving +# all the data in a file forward $n bytes +# ($n may be negative) +# The result should be at least $n. +sub _bufsize { + my $n = shift; + return 8192 if $n <= 0; + my $b = $n & ~8191; + $b += 8192 if $n & 8191; + $b; +} + +################################################################ +# +# Miscellaneous public methods +# + +# Lock the file +sub flock { + my ($self, $op) = @_; + unless (@_ <= 3) { + my $pack = ref $self; + croak "Usage: $pack\->flock([OPERATION])"; + } + my $fh = $self->{fh}; + $op = LOCK_EX unless defined $op; + my $locked = flock $fh, $op; + + if ($locked && ($op & (LOCK_EX | LOCK_SH))) { + # If you're locking the file, then presumably it's because + # there might have been a write access by another process. + # In that case, the read cache contents and the offsets table + # might be invalid, so discard them. 20030508 + $self->{offsets} = [0]; + $self->{cache}->empty; + } + + $locked; +} + +# Get/set autochomp option +sub autochomp { + my $self = shift; + if (@_) { + my $old = $self->{autochomp}; + $self->{autochomp} = shift; + $old; + } else { + $self->{autochomp}; + } +} + +# Get offset table entries; returns offset of nth record +sub offset { + my ($self, $n) = @_; + + if ($#{$self->{offsets}} < $n) { + return if $self->{eof}; # request for record beyond the end of file + my $o = $self->_fill_offsets_to($n); + # If it's still undefined, there is no such record, so return 'undef' + return unless defined $o; + } + + $self->{offsets}[$n]; +} + +sub discard_offsets { + my $self = shift; + $self->{offsets} = [0]; +} + +################################################################ +# +# Matters related to deferred writing +# + +# Defer writes +sub defer { + my $self = shift; + $self->_stop_autodeferring; + @{$self->{ad_history}} = (); + $self->{defer} = 1; +} + +# Flush deferred writes +# +# This could be better optimized to write the file in one pass, instead +# of one pass per block of records. But that will require modifications +# to _twrite, so I should have a good _twrite test suite first. +sub flush { + my $self = shift; + + $self->_flush; + $self->{defer} = 0; +} + +sub _old_flush { + my $self = shift; + my @writable = sort {$a<=>$b} (keys %{$self->{deferred}}); + + while (@writable) { + # gather all consecutive records from the front of @writable + my $first_rec = shift @writable; + my $last_rec = $first_rec+1; + ++$last_rec, shift @writable while @writable && $last_rec == $writable[0]; + --$last_rec; + $self->_fill_offsets_to($last_rec); + $self->_extend_file_to($last_rec); + $self->_splice($first_rec, $last_rec-$first_rec+1, + @{$self->{deferred}}{$first_rec .. $last_rec}); + } + + $self->_discard; # clear out defered-write-cache +} + +sub _flush { + my $self = shift; + my @writable = sort {$a<=>$b} (keys %{$self->{deferred}}); + my @args; + my @adjust; + + while (@writable) { + # gather all consecutive records from the front of @writable + my $first_rec = shift @writable; + my $last_rec = $first_rec+1; + ++$last_rec, shift @writable while @writable && $last_rec == $writable[0]; + --$last_rec; + my $end = $self->_fill_offsets_to($last_rec+1); + if (not defined $end) { + $self->_extend_file_to($last_rec); + $end = $self->{offsets}[$last_rec]; + } + my ($start) = $self->{offsets}[$first_rec]; + push @args, + join("", @{$self->{deferred}}{$first_rec .. $last_rec}), # data + $start, # position + $end-$start; # length + push @adjust, [$first_rec, # starting at this position... + $last_rec-$first_rec+1, # this many records... + # are replaced with these... + @{$self->{deferred}}{$first_rec .. $last_rec}, + ]; + } + + $self->_mtwrite(@args); # write multiple record groups + $self->_discard; # clear out defered-write-cache + $self->_oadjust(@adjust); +} + +# Discard deferred writes and disable future deferred writes +sub discard { + my $self = shift; + $self->_discard; + $self->{defer} = 0; +} + +# Discard deferred writes, but retain old deferred writing mode +sub _discard { + my $self = shift; + %{$self->{deferred}} = (); + $self->{deferred_s} = 0; + $self->{deferred_max} = -1; + $self->{cache}->set_limit($self->{memory}); +} + +# Deferred writing is enabled, either explicitly ($self->{defer}) +# or automatically ($self->{autodeferring}) +sub _is_deferring { + my $self = shift; + $self->{defer} || $self->{autodeferring}; +} + +# The largest record number of any deferred record +sub _defer_max { + my $self = shift; + return $self->{deferred_max} if defined $self->{deferred_max}; + my $max = -1; + for my $key (keys %{$self->{deferred}}) { + $max = $key if $key > $max; + } + $self->{deferred_max} = $max; + $max; +} + +################################################################ +# +# Matters related to autodeferment +# + +# Get/set autodefer option +sub autodefer { + my $self = shift; + if (@_) { + my $old = $self->{autodefer}; + $self->{autodefer} = shift; + if ($old) { + $self->_stop_autodeferring; + @{$self->{ad_history}} = (); + } + $old; + } else { + $self->{autodefer}; + } +} + +# The user is trying to store record #$n Record that in the history, +# and then enable (or disable) autodeferment if that seems useful. +# Note that it's OK for $n to be a non-number, as long as the function +# is prepared to deal with that. Nobody else looks at the ad_history. +# +# Now, what does the ad_history mean, and what is this function doing? +# Essentially, the idea is to enable autodeferring when we see that the +# user has made three consecutive STORE calls to three consecutive records. +# ("Three" is actually ->{autodefer_threshhold}.) +# A STORE call for record #$n inserts $n into the autodefer history, +# and if the history contains three consecutive records, we enable +# autodeferment. An ad_history of [X, Y] means that the most recent +# STOREs were for records X, X+1, ..., Y, in that order. +# +# Inserting a nonconsecutive number erases the history and starts over. +# +# Performing a special operation like SPLICE erases the history. +# +# There's one special case: CLEAR means that CLEAR was just called. +# In this case, we prime the history with [-2, -1] so that if the next +# write is for record 0, autodeferring goes on immediately. This is for +# the common special case of "@a = (...)". +# +sub _annotate_ad_history { + my ($self, $n) = @_; + return unless $self->{autodefer}; # feature is disabled + return if $self->{defer}; # already in explicit defer mode + return unless $self->{offsets}[-1] >= $self->{autodefer_filelen_threshhold}; + + local *H = $self->{ad_history}; + if ($n eq 'CLEAR') { + @H = (-2, -1); # prime the history with fake records + $self->_stop_autodeferring; + } elsif ($n =~ /^\d+$/) { + if (@H == 0) { + @H = ($n, $n); + } else { # @H == 2 + if ($H[1] == $n-1) { # another consecutive record + $H[1]++; + if ($H[1] - $H[0] + 1 >= $self->{autodefer_threshhold}) { + $self->{autodeferring} = 1; + } + } else { # nonconsecutive- erase and start over + @H = ($n, $n); + $self->_stop_autodeferring; + } + } + } else { # SPLICE or STORESIZE or some such + @H = (); + $self->_stop_autodeferring; + } +} + +# If autodeferring was enabled, cut it out and discard the history +sub _stop_autodeferring { + my $self = shift; + if ($self->{autodeferring}) { + $self->_flush; + } + $self->{autodeferring} = 0; +} + +################################################################ + + +# This is NOT a method. It is here for two reasons: +# 1. To factor a fairly complicated block out of the constructor +# 2. To provide access for the test suite, which need to be sure +# files are being written properly. +sub _default_recsep { + my $recsep = $/; + if ($^O eq 'MSWin32') { # Dos too? + # Windows users expect files to be terminated with \r\n + # But $/ is set to \n instead + # Note that this also transforms \n\n into \r\n\r\n. + # That is a feature. + $recsep =~ s/\n/\r\n/g; + } + $recsep; +} + +# Utility function for _check_integrity +sub _ci_warn { + my $msg = shift; + $msg =~ s/\n/\\n/g; + $msg =~ s/\r/\\r/g; + print "# $msg\n"; +} + +# Given a file, make sure the cache is consistent with the +# file contents and the internal data structures are consistent with +# each other. Returns true if everything checks out, false if not +# +# The $file argument is no longer used. It is retained for compatibility +# with the existing test suite. +sub _check_integrity { + my ($self, $file, $warn) = @_; + my $rsl = $self->{recseplen}; + my $rs = $self->{recsep}; + my $good = 1; + local *_; # local $_ does not work here + local $DIAGNOSTIC = 1; + + if (not defined $rs) { + _ci_warn("recsep is undef!"); + $good = 0; + } elsif ($rs eq "") { + _ci_warn("recsep is empty!"); + $good = 0; + } elsif ($rsl != length $rs) { + my $ln = length $rs; + _ci_warn("recsep <$rs> has length $ln, should be $rsl"); + $good = 0; + } + + if (not defined $self->{offsets}[0]) { + _ci_warn("offset 0 is missing!"); + $good = 0; + + } elsif ($self->{offsets}[0] != 0) { + _ci_warn("rec 0: offset <$self->{offsets}[0]> s/b 0!"); + $good = 0; + } + + my $cached = 0; + { + local *F = $self->{fh}; + seek F, 0, SEEK_SET; + local $. = 0; + local $/ = $rs; + + while (<F>) { + my $n = $. - 1; + my $cached = $self->{cache}->_produce($n); + my $offset = $self->{offsets}[$.]; + my $ao = tell F; + if (defined $offset && $offset != $ao) { + _ci_warn("rec $n: offset <$offset> actual <$ao>"); + $good = 0; + } + if (defined $cached && $_ ne $cached && ! $self->{deferred}{$n}) { + $good = 0; + _ci_warn("rec $n: cached <$cached> actual <$_>"); + } + if (defined $cached && substr($cached, -$rsl) ne $rs) { + $good = 0; + _ci_warn("rec $n in the cache is missing the record separator"); + } + if (! defined $offset && $self->{eof}) { + $good = 0; + _ci_warn("The offset table was marked complete, but it is missing element $."); + } + } + if (@{$self->{offsets}} > $.+1) { + $good = 0; + my $n = @{$self->{offsets}}; + _ci_warn("The offset table has $n items, but the file has only $."); + } + + my $deferring = $self->_is_deferring; + for my $n ($self->{cache}->ckeys) { + my $r = $self->{cache}->_produce($n); + $cached += length($r); + next if $n+1 <= $.; # checked this already + _ci_warn("spurious caching of record $n"); + $good = 0; + } + my $b = $self->{cache}->bytes; + if ($cached != $b) { + _ci_warn("cache size is $b, should be $cached"); + $good = 0; + } + } + + # That cache has its own set of tests + $good = 0 unless $self->{cache}->_check_integrity; + + # Now let's check the deferbuffer + # Unless deferred writing is enabled, it should be empty + if (! $self->_is_deferring && %{$self->{deferred}}) { + _ci_warn("deferred writing disabled, but deferbuffer nonempty"); + $good = 0; + } + + # Any record in the deferbuffer should *not* be present in the readcache + my $deferred_s = 0; + while (my ($n, $r) = each %{$self->{deferred}}) { + $deferred_s += length($r); + if (defined $self->{cache}->_produce($n)) { + _ci_warn("record $n is in the deferbuffer *and* the readcache"); + $good = 0; + } + if (substr($r, -$rsl) ne $rs) { + _ci_warn("rec $n in the deferbuffer is missing the record separator"); + $good = 0; + } + } + + # Total size of deferbuffer should match internal total + if ($deferred_s != $self->{deferred_s}) { + _ci_warn("buffer size is $self->{deferred_s}, should be $deferred_s"); + $good = 0; + } + + # Total size of deferbuffer should not exceed the specified limit + if ($deferred_s > $self->{dw_size}) { + _ci_warn("buffer size is $self->{deferred_s} which exceeds the limit of $self->{dw_size}"); + $good = 0; + } + + # Total size of cached data should not exceed the specified limit + if ($deferred_s + $cached > $self->{memory}) { + my $total = $deferred_s + $cached; + _ci_warn("total stored data size is $total which exceeds the limit of $self->{memory}"); + $good = 0; + } + + # Stuff related to autodeferment + if (!$self->{autodefer} && @{$self->{ad_history}}) { + _ci_warn("autodefer is disabled, but ad_history is nonempty"); + $good = 0; + } + if ($self->{autodeferring} && $self->{defer}) { + _ci_warn("both autodeferring and explicit deferring are active"); + $good = 0; + } + if (@{$self->{ad_history}} == 0) { + # That's OK, no additional tests required + } elsif (@{$self->{ad_history}} == 2) { + my @non_number = grep !/^-?\d+$/, @{$self->{ad_history}}; + if (@non_number) { + my $msg; + { local $" = ')('; + $msg = "ad_history contains non-numbers (@{$self->{ad_history}})"; + } + _ci_warn($msg); + $good = 0; + } elsif ($self->{ad_history}[1] < $self->{ad_history}[0]) { + _ci_warn("ad_history has nonsensical values @{$self->{ad_history}}"); + $good = 0; + } + } else { + _ci_warn("ad_history has bad length <@{$self->{ad_history}}>"); + $good = 0; + } + + $good; +} + +################################################################ +# +# Tie::File::Cache +# +# Read cache + +package Tie::File::Cache; +$Tie::File::Cache::VERSION = $Tie::File::VERSION; +use Carp ':DEFAULT', 'confess'; + +sub HEAP () { 0 } +sub HASH () { 1 } +sub MAX () { 2 } +sub BYTES() { 3 } +#sub STAT () { 4 } # Array with request statistics for each record +#sub MISS () { 5 } # Total number of cache misses +#sub REQ () { 6 } # Total number of cache requests +use strict 'vars'; + +sub new { + my ($pack, $max) = @_; + local *_; + croak "missing argument to ->new" unless defined $max; + my $self = []; + bless $self => $pack; + @$self = (Tie::File::Heap->new($self), {}, $max, 0); + $self; +} + +sub adj_limit { + my ($self, $n) = @_; + $self->[MAX] += $n; +} + +sub set_limit { + my ($self, $n) = @_; + $self->[MAX] = $n; +} + +# For internal use only +# Will be called by the heap structure to notify us that a certain +# piece of data has moved from one heap element to another. +# $k is the hash key of the item +# $n is the new index into the heap at which it is stored +# If $n is undefined, the item has been removed from the heap. +sub _heap_move { + my ($self, $k, $n) = @_; + if (defined $n) { + $self->[HASH]{$k} = $n; + } else { + delete $self->[HASH]{$k}; + } +} + +sub insert { + my ($self, $key, $val) = @_; + local *_; + croak "missing argument to ->insert" unless defined $key; + unless (defined $self->[MAX]) { + confess "undefined max" ; + } + confess "undefined val" unless defined $val; + return if length($val) > $self->[MAX]; + +# if ($self->[STAT]) { +# $self->[STAT][$key] = 1; +# return; +# } + + my $oldnode = $self->[HASH]{$key}; + if (defined $oldnode) { + my $oldval = $self->[HEAP]->set_val($oldnode, $val); + $self->[BYTES] -= length($oldval); + } else { + $self->[HEAP]->insert($key, $val); + } + $self->[BYTES] += length($val); + $self->flush if $self->[BYTES] > $self->[MAX]; +} + +sub expire { + my $self = shift; + my $old_data = $self->[HEAP]->popheap; + return unless defined $old_data; + $self->[BYTES] -= length $old_data; + $old_data; +} + +sub remove { + my ($self, @keys) = @_; + my @result; + +# if ($self->[STAT]) { +# for my $key (@keys) { +# $self->[STAT][$key] = 0; +# } +# return; +# } + + for my $key (@keys) { + next unless exists $self->[HASH]{$key}; + my $old_data = $self->[HEAP]->remove($self->[HASH]{$key}); + $self->[BYTES] -= length $old_data; + push @result, $old_data; + } + @result; +} + +sub lookup { + my ($self, $key) = @_; + local *_; + croak "missing argument to ->lookup" unless defined $key; + +# if ($self->[STAT]) { +# $self->[MISS]++ if $self->[STAT][$key]++ == 0; +# $self->[REQ]++; +# my $hit_rate = 1 - $self->[MISS] / $self->[REQ]; +# # Do some testing to determine this threshhold +# $#$self = STAT - 1 if $hit_rate > 0.20; +# } + + if (exists $self->[HASH]{$key}) { + $self->[HEAP]->lookup($self->[HASH]{$key}); + } else { + return; + } +} + +# For internal use only +sub _produce { + my ($self, $key) = @_; + my $loc = $self->[HASH]{$key}; + return unless defined $loc; + $self->[HEAP][$loc][2]; +} + +# For internal use only +sub _promote { + my ($self, $key) = @_; + $self->[HEAP]->promote($self->[HASH]{$key}); +} + +sub empty { + my ($self) = @_; + %{$self->[HASH]} = (); + $self->[BYTES] = 0; + $self->[HEAP]->empty; +# @{$self->[STAT]} = (); +# $self->[MISS] = 0; +# $self->[REQ] = 0; +} + +sub is_empty { + my ($self) = @_; + keys %{$self->[HASH]} == 0; +} + +sub update { + my ($self, $key, $val) = @_; + local *_; + croak "missing argument to ->update" unless defined $key; + if (length($val) > $self->[MAX]) { + my ($oldval) = $self->remove($key); + $self->[BYTES] -= length($oldval) if defined $oldval; + } elsif (exists $self->[HASH]{$key}) { + my $oldval = $self->[HEAP]->set_val($self->[HASH]{$key}, $val); + $self->[BYTES] += length($val); + $self->[BYTES] -= length($oldval) if defined $oldval; + } else { + $self->[HEAP]->insert($key, $val); + $self->[BYTES] += length($val); + } + $self->flush; +} + +sub rekey { + my ($self, $okeys, $nkeys) = @_; + local *_; + my %map; + @map{@$okeys} = @$nkeys; + croak "missing argument to ->rekey" unless defined $nkeys; + croak "length mismatch in ->rekey arguments" unless @$nkeys == @$okeys; + my %adjusted; # map new keys to heap indices + # You should be able to cut this to one loop TODO XXX + for (0 .. $#$okeys) { + $adjusted{$nkeys->[$_]} = delete $self->[HASH]{$okeys->[$_]}; + } + while (my ($nk, $ix) = each %adjusted) { + # @{$self->[HASH]}{keys %adjusted} = values %adjusted; + $self->[HEAP]->rekey($ix, $nk); + $self->[HASH]{$nk} = $ix; + } +} + +sub ckeys { + my $self = shift; + my @a = keys %{$self->[HASH]}; + @a; +} + +# Return total amount of cached data +sub bytes { + my $self = shift; + $self->[BYTES]; +} + +# Expire oldest item from cache until cache size is smaller than $max +sub reduce_size_to { + my ($self, $max) = @_; + until ($self->[BYTES] <= $max) { + # Note that Tie::File::Cache::expire has been inlined here + my $old_data = $self->[HEAP]->popheap; + return unless defined $old_data; + $self->[BYTES] -= length $old_data; + } +} + +# Why not just $self->reduce_size_to($self->[MAX])? +# Try this when things stabilize TODO XXX +# If the cache is too full, expire the oldest records +sub flush { + my $self = shift; + $self->reduce_size_to($self->[MAX]) if $self->[BYTES] > $self->[MAX]; +} + +# For internal use only +sub _produce_lru { + my $self = shift; + $self->[HEAP]->expire_order; +} + +BEGIN { *_ci_warn = \&Tie::File::_ci_warn } + +sub _check_integrity { # For CACHE + my $self = shift; + my $good = 1; + + # Test HEAP + $self->[HEAP]->_check_integrity or $good = 0; + + # Test HASH + my $bytes = 0; + for my $k (keys %{$self->[HASH]}) { + if ($k ne '0' && $k !~ /^[1-9][0-9]*$/) { + $good = 0; + _ci_warn "Cache hash key <$k> is non-numeric"; + } + + my $h = $self->[HASH]{$k}; + if (! defined $h) { + $good = 0; + _ci_warn "Heap index number for key $k is undefined"; + } elsif ($h == 0) { + $good = 0; + _ci_warn "Heap index number for key $k is zero"; + } else { + my $j = $self->[HEAP][$h]; + if (! defined $j) { + $good = 0; + _ci_warn "Heap contents key $k (=> $h) are undefined"; + } else { + $bytes += length($j->[2]); + if ($k ne $j->[1]) { + $good = 0; + _ci_warn "Heap contents key $k (=> $h) is $j->[1], should be $k"; + } + } + } + } + + # Test BYTES + if ($bytes != $self->[BYTES]) { + $good = 0; + _ci_warn "Total data in cache is $bytes, expected $self->[BYTES]"; + } + + # Test MAX + if ($bytes > $self->[MAX]) { + $good = 0; + _ci_warn "Total data in cache is $bytes, exceeds maximum $self->[MAX]"; + } + + return $good; +} + +sub delink { + my $self = shift; + $self->[HEAP] = undef; # Bye bye heap +} + +################################################################ +# +# Tie::File::Heap +# +# Heap data structure for use by cache LRU routines + +package Tie::File::Heap; +use Carp ':DEFAULT', 'confess'; +$Tie::File::Heap::VERSION = $Tie::File::Cache::VERSION; +sub SEQ () { 0 }; +sub KEY () { 1 }; +sub DAT () { 2 }; + +sub new { + my ($pack, $cache) = @_; + die "$pack: Parent cache object $cache does not support _heap_move method" + unless eval { $cache->can('_heap_move') }; + my $self = [[0,$cache,0]]; + bless $self => $pack; +} + +# Allocate a new sequence number, larger than all previously allocated numbers +sub _nseq { + my $self = shift; + $self->[0][0]++; +} + +sub _cache { + my $self = shift; + $self->[0][1]; +} + +sub _nelts { + my $self = shift; + $self->[0][2]; +} + +sub _nelts_inc { + my $self = shift; + ++$self->[0][2]; +} + +sub _nelts_dec { + my $self = shift; + --$self->[0][2]; +} + +sub is_empty { + my $self = shift; + $self->_nelts == 0; +} + +sub empty { + my $self = shift; + $#$self = 0; + $self->[0][2] = 0; + $self->[0][0] = 0; # might as well reset the sequence numbers +} + +# notify the parent cache object that we moved something +sub _heap_move { + my $self = shift; + $self->_cache->_heap_move(@_); +} + +# Insert a piece of data into the heap with the indicated sequence number. +# The item with the smallest sequence number is always at the top. +# If no sequence number is specified, allocate a new one and insert the +# item at the bottom. +sub insert { + my ($self, $key, $data, $seq) = @_; + $seq = $self->_nseq unless defined $seq; + $self->_insert_new([$seq, $key, $data]); +} + +# Insert a new, fresh item at the bottom of the heap +sub _insert_new { + my ($self, $item) = @_; + my $i = @$self; + $i = int($i/2) until defined $self->[$i/2]; + $self->[$i] = $item; + $self->[0][1]->_heap_move($self->[$i][KEY], $i); + $self->_nelts_inc; +} + +# Insert [$data, $seq] pair at or below item $i in the heap. +# If $i is omitted, default to 1 (the top element.) +sub _insert { + my ($self, $item, $i) = @_; +# $self->_check_loc($i) if defined $i; + $i = 1 unless defined $i; + until (! defined $self->[$i]) { + if ($self->[$i][SEQ] > $item->[SEQ]) { # inserted item is older + ($self->[$i], $item) = ($item, $self->[$i]); + $self->[0][1]->_heap_move($self->[$i][KEY], $i); + } + # If either is undefined, go that way. Otherwise, choose at random + my $dir; + $dir = 0 if !defined $self->[2*$i]; + $dir = 1 if !defined $self->[2*$i+1]; + $dir = int(rand(2)) unless defined $dir; + $i = 2*$i + $dir; + } + $self->[$i] = $item; + $self->[0][1]->_heap_move($self->[$i][KEY], $i); + $self->_nelts_inc; +} + +# Remove the item at node $i from the heap, moving child items upwards. +# The item with the smallest sequence number is always at the top. +# Moving items upwards maintains this condition. +# Return the removed item. Return undef if there was no item at node $i. +sub remove { + my ($self, $i) = @_; + $i = 1 unless defined $i; + my $top = $self->[$i]; + return unless defined $top; + while (1) { + my $ii; + my ($L, $R) = (2*$i, 2*$i+1); + + # If either is undefined, go the other way. + # Otherwise, go towards the smallest. + last unless defined $self->[$L] || defined $self->[$R]; + $ii = $R if not defined $self->[$L]; + $ii = $L if not defined $self->[$R]; + unless (defined $ii) { + $ii = $self->[$L][SEQ] < $self->[$R][SEQ] ? $L : $R; + } + + $self->[$i] = $self->[$ii]; # Promote child to fill vacated spot + $self->[0][1]->_heap_move($self->[$i][KEY], $i); + $i = $ii; # Fill new vacated spot + } + $self->[0][1]->_heap_move($top->[KEY], undef); + undef $self->[$i]; + $self->_nelts_dec; + return $top->[DAT]; +} + +sub popheap { + my $self = shift; + $self->remove(1); +} + +# set the sequence number of the indicated item to a higher number +# than any other item in the heap, and bubble the item down to the +# bottom. +sub promote { + my ($self, $n) = @_; +# $self->_check_loc($n); + $self->[$n][SEQ] = $self->_nseq; + my $i = $n; + while (1) { + my ($L, $R) = (2*$i, 2*$i+1); + my $dir; + last unless defined $self->[$L] || defined $self->[$R]; + $dir = $R unless defined $self->[$L]; + $dir = $L unless defined $self->[$R]; + unless (defined $dir) { + $dir = $self->[$L][SEQ] < $self->[$R][SEQ] ? $L : $R; + } + @{$self}[$i, $dir] = @{$self}[$dir, $i]; + for ($i, $dir) { + $self->[0][1]->_heap_move($self->[$_][KEY], $_) if defined $self->[$_]; + } + $i = $dir; + } +} + +# Return item $n from the heap, promoting its LRU status +sub lookup { + my ($self, $n) = @_; +# $self->_check_loc($n); + my $val = $self->[$n]; + $self->promote($n); + $val->[DAT]; +} + + +# Assign a new value for node $n, promoting it to the bottom of the heap +sub set_val { + my ($self, $n, $val) = @_; +# $self->_check_loc($n); + my $oval = $self->[$n][DAT]; + $self->[$n][DAT] = $val; + $self->promote($n); + return $oval; +} + +# The hask key has changed for an item; +# alter the heap's record of the hash key +sub rekey { + my ($self, $n, $new_key) = @_; +# $self->_check_loc($n); + $self->[$n][KEY] = $new_key; +} + +sub _check_loc { + my ($self, $n) = @_; + unless (1 || defined $self->[$n]) { + confess "_check_loc($n) failed"; + } +} + +BEGIN { *_ci_warn = \&Tie::File::_ci_warn } + +sub _check_integrity { + my $self = shift; + my $good = 1; + my %seq; + + unless (eval {$self->[0][1]->isa("Tie::File::Cache")}) { + _ci_warn "Element 0 of heap corrupt"; + $good = 0; + } + $good = 0 unless $self->_satisfies_heap_condition(1); + for my $i (2 .. $#{$self}) { + my $p = int($i/2); # index of parent node + if (defined $self->[$i] && ! defined $self->[$p]) { + _ci_warn "Element $i of heap defined, but parent $p isn't"; + $good = 0; + } + + if (defined $self->[$i]) { + if ($seq{$self->[$i][SEQ]}) { + my $seq = $self->[$i][SEQ]; + _ci_warn "Nodes $i and $seq{$seq} both have SEQ=$seq"; + $good = 0; + } else { + $seq{$self->[$i][SEQ]} = $i; + } + } + } + + return $good; +} + +sub _satisfies_heap_condition { + my $self = shift; + my $n = shift || 1; + my $good = 1; + for (0, 1) { + my $c = $n*2 + $_; + next unless defined $self->[$c]; + if ($self->[$n][SEQ] >= $self->[$c]) { + _ci_warn "Node $n of heap does not predate node $c"; + $good = 0 ; + } + $good = 0 unless $self->_satisfies_heap_condition($c); + } + return $good; +} + +# Return a list of all the values, sorted by expiration order +sub expire_order { + my $self = shift; + my @nodes = sort {$a->[SEQ] <=> $b->[SEQ]} $self->_nodes; + map { $_->[KEY] } @nodes; +} + +sub _nodes { + my $self = shift; + my $i = shift || 1; + return unless defined $self->[$i]; + ($self->[$i], $self->_nodes($i*2), $self->_nodes($i*2+1)); +} + +"Cogito, ergo sum."; # don't forget to return a true value from the file + +__END__ + +=head1 NAME + +Tie::File - Access the lines of a disk file via a Perl array + +=head1 SYNOPSIS + + # This file documents Tie::File version 0.97 + use Tie::File; + + tie @array, 'Tie::File', filename or die ...; + + $array[13] = 'blah'; # line 13 of the file is now 'blah' + print $array[42]; # display line 42 of the file + + $n_recs = @array; # how many records are in the file? + $#array -= 2; # chop two records off the end + + + for (@array) { + s/PERL/Perl/g; # Replace PERL with Perl everywhere in the file + } + + # These are just like regular push, pop, unshift, shift, and splice + # Except that they modify the file in the way you would expect + + push @array, new recs...; + my $r1 = pop @array; + unshift @array, new recs...; + my $r2 = shift @array; + @old_recs = splice @array, 3, 7, new recs...; + + untie @array; # all finished + + +=head1 DESCRIPTION + +C<Tie::File> represents a regular text file as a Perl array. Each +element in the array corresponds to a record in the file. The first +line of the file is element 0 of the array; the second line is element +1, and so on. + +The file is I<not> loaded into memory, so this will work even for +gigantic files. + +Changes to the array are reflected in the file immediately. + +Lazy people and beginners may now stop reading the manual. + +=head2 C<recsep> + +What is a 'record'? By default, the meaning is the same as for the +C<E<lt>...E<gt>> operator: It's a string terminated by C<$/>, which is +probably C<"\n">. (Minor exception: on DOS and Win32 systems, a +'record' is a string terminated by C<"\r\n">.) You may change the +definition of "record" by supplying the C<recsep> option in the C<tie> +call: + + tie @array, 'Tie::File', $file, recsep => 'es'; + +This says that records are delimited by the string C<es>. If the file +contained the following data: + + Curse these pesky flies!\n + +then the C<@array> would appear to have four elements: + + "Curse th" + "e p" + "ky fli" + "!\n" + +An undefined value is not permitted as a record separator. Perl's +special "paragraph mode" semantics (E<agrave> la C<$/ = "">) are not +emulated. + +Records read from the tied array do not have the record separator +string on the end; this is to allow + + $array[17] .= "extra"; + +to work as expected. + +(See L<"autochomp">, below.) Records stored into the array will have +the record separator string appended before they are written to the +file, if they don't have one already. For example, if the record +separator string is C<"\n">, then the following two lines do exactly +the same thing: + + $array[17] = "Cherry pie"; + $array[17] = "Cherry pie\n"; + +The result is that the contents of line 17 of the file will be +replaced with "Cherry pie"; a newline character will separate line 17 +from line 18. This means that this code will do nothing: + + chomp $array[17]; + +Because the C<chomp>ed value will have the separator reattached when +it is written back to the file. There is no way to create a file +whose trailing record separator string is missing. + +Inserting records that I<contain> the record separator string is not +supported by this module. It will probably produce a reasonable +result, but what this result will be may change in a future version. +Use 'splice' to insert records or to replace one record with several. + +=head2 C<autochomp> + +Normally, array elements have the record separator removed, so that if +the file contains the text + + Gold + Frankincense + Myrrh + +the tied array will appear to contain C<("Gold", "Frankincense", +"Myrrh")>. If you set C<autochomp> to a false value, the record +separator will not be removed. If the file above was tied with + + tie @gifts, "Tie::File", $gifts, autochomp => 0; + +then the array C<@gifts> would appear to contain C<("Gold\n", +"Frankincense\n", "Myrrh\n")>, or (on Win32 systems) C<("Gold\r\n", +"Frankincense\r\n", "Myrrh\r\n")>. + +=head2 C<mode> + +Normally, the specified file will be opened for read and write access, +and will be created if it does not exist. (That is, the flags +C<O_RDWR | O_CREAT> are supplied in the C<open> call.) If you want to +change this, you may supply alternative flags in the C<mode> option. +See L<Fcntl> for a listing of available flags. +For example: + + # open the file if it exists, but fail if it does not exist + use Fcntl 'O_RDWR'; + tie @array, 'Tie::File', $file, mode => O_RDWR; + + # create the file if it does not exist + use Fcntl 'O_RDWR', 'O_CREAT'; + tie @array, 'Tie::File', $file, mode => O_RDWR | O_CREAT; + + # open an existing file in read-only mode + use Fcntl 'O_RDONLY'; + tie @array, 'Tie::File', $file, mode => O_RDONLY; + +Opening the data file in write-only or append mode is not supported. + +=head2 C<memory> + +This is an upper limit on the amount of memory that C<Tie::File> will +consume at any time while managing the file. This is used for two +things: managing the I<read cache> and managing the I<deferred write +buffer>. + +Records read in from the file are cached, to avoid having to re-read +them repeatedly. If you read the same record twice, the first time it +will be stored in memory, and the second time it will be fetched from +the I<read cache>. The amount of data in the read cache will not +exceed the value you specified for C<memory>. If C<Tie::File> wants +to cache a new record, but the read cache is full, it will make room +by expiring the least-recently visited records from the read cache. + +The default memory limit is 2Mib. You can adjust the maximum read +cache size by supplying the C<memory> option. The argument is the +desired cache size, in bytes. + + # I have a lot of memory, so use a large cache to speed up access + tie @array, 'Tie::File', $file, memory => 20_000_000; + +Setting the memory limit to 0 will inhibit caching; records will be +fetched from disk every time you examine them. + +The C<memory> value is not an absolute or exact limit on the memory +used. C<Tie::File> objects contains some structures besides the read +cache and the deferred write buffer, whose sizes are not charged +against C<memory>. + +The cache itself consumes about 310 bytes per cached record, so if +your file has many short records, you may want to decrease the cache +memory limit, or else the cache overhead may exceed the size of the +cached data. + + +=head2 C<dw_size> + +(This is an advanced feature. Skip this section on first reading.) + +If you use deferred writing (See L<"Deferred Writing">, below) then +data you write into the array will not be written directly to the +file; instead, it will be saved in the I<deferred write buffer> to be +written out later. Data in the deferred write buffer is also charged +against the memory limit you set with the C<memory> option. + +You may set the C<dw_size> option to limit the amount of data that can +be saved in the deferred write buffer. This limit may not exceed the +total memory limit. For example, if you set C<dw_size> to 1000 and +C<memory> to 2500, that means that no more than 1000 bytes of deferred +writes will be saved up. The space available for the read cache will +vary, but it will always be at least 1500 bytes (if the deferred write +buffer is full) and it could grow as large as 2500 bytes (if the +deferred write buffer is empty.) + +If you don't specify a C<dw_size>, it defaults to the entire memory +limit. + +=head2 Option Format + +C<-mode> is a synonym for C<mode>. C<-recsep> is a synonym for +C<recsep>. C<-memory> is a synonym for C<memory>. You get the +idea. + +=head1 Public Methods + +The C<tie> call returns an object, say C<$o>. You may call + + $rec = $o->FETCH($n); + $o->STORE($n, $rec); + +to fetch or store the record at line C<$n>, respectively; similarly +the other tied array methods. (See L<perltie> for details.) You may +also call the following methods on this object: + +=head2 C<flock> + + $o->flock(MODE) + +will lock the tied file. C<MODE> has the same meaning as the second +argument to the Perl built-in C<flock> function; for example +C<LOCK_SH> or C<LOCK_EX | LOCK_NB>. (These constants are provided by +the C<use Fcntl ':flock'> declaration.) + +C<MODE> is optional; the default is C<LOCK_EX>. + +C<Tie::File> maintains an internal table of the byte offset of each +record it has seen in the file. + +When you use C<flock> to lock the file, C<Tie::File> assumes that the +read cache is no longer trustworthy, because another process might +have modified the file since the last time it was read. Therefore, a +successful call to C<flock> discards the contents of the read cache +and the internal record offset table. + +C<Tie::File> promises that the following sequence of operations will +be safe: + + my $o = tie @array, "Tie::File", $filename; + $o->flock; + +In particular, C<Tie::File> will I<not> read or write the file during +the C<tie> call. (Exception: Using C<mode =E<gt> O_TRUNC> will, of +course, erase the file during the C<tie> call. If you want to do this +safely, then open the file without C<O_TRUNC>, lock the file, and use +C<@array = ()>.) + +The best way to unlock a file is to discard the object and untie the +array. It is probably unsafe to unlock the file without also untying +it, because if you do, changes may remain unwritten inside the object. +That is why there is no shortcut for unlocking. If you really want to +unlock the file prematurely, you know what to do; if you don't know +what to do, then don't do it. + +All the usual warnings about file locking apply here. In particular, +note that file locking in Perl is B<advisory>, which means that +holding a lock will not prevent anyone else from reading, writing, or +erasing the file; it only prevents them from getting another lock at +the same time. Locks are analogous to green traffic lights: If you +have a green light, that does not prevent the idiot coming the other +way from plowing into you sideways; it merely guarantees to you that +the idiot does not also have a green light at the same time. + +=head2 C<autochomp> + + my $old_value = $o->autochomp(0); # disable autochomp option + my $old_value = $o->autochomp(1); # enable autochomp option + + my $ac = $o->autochomp(); # recover current value + +See L<"autochomp">, above. + +=head2 C<defer>, C<flush>, C<discard>, and C<autodefer> + +See L<"Deferred Writing">, below. + +=head2 C<offset> + + $off = $o->offset($n); + +This method returns the byte offset of the start of the C<$n>th record +in the file. If there is no such record, it returns an undefined +value. + +=head1 Tying to an already-opened filehandle + +If C<$fh> is a filehandle, such as is returned by C<IO::File> or one +of the other C<IO> modules, you may use: + + tie @array, 'Tie::File', $fh, ...; + +Similarly if you opened that handle C<FH> with regular C<open> or +C<sysopen>, you may use: + + tie @array, 'Tie::File', \*FH, ...; + +Handles that were opened write-only won't work. Handles that were +opened read-only will work as long as you don't try to modify the +array. Handles must be attached to seekable sources of data---that +means no pipes or sockets. If C<Tie::File> can detect that you +supplied a non-seekable handle, the C<tie> call will throw an +exception. (On Unix systems, it can detect this.) + +Note that Tie::File will only close any filehandles that it opened +internally. If you passed it a filehandle as above, you "own" the +filehandle, and are responsible for closing it after you have untied +the @array. + +=head1 Deferred Writing + +(This is an advanced feature. Skip this section on first reading.) + +Normally, modifying a C<Tie::File> array writes to the underlying file +immediately. Every assignment like C<$a[3] = ...> rewrites as much of +the file as is necessary; typically, everything from line 3 through +the end will need to be rewritten. This is the simplest and most +transparent behavior. Performance even for large files is reasonably +good. + +However, under some circumstances, this behavior may be excessively +slow. For example, suppose you have a million-record file, and you +want to do: + + for (@FILE) { + $_ = "> $_"; + } + +The first time through the loop, you will rewrite the entire file, +from line 0 through the end. The second time through the loop, you +will rewrite the entire file from line 1 through the end. The third +time through the loop, you will rewrite the entire file from line 2 to +the end. And so on. + +If the performance in such cases is unacceptable, you may defer the +actual writing, and then have it done all at once. The following loop +will perform much better for large files: + + (tied @a)->defer; + for (@a) { + $_ = "> $_"; + } + (tied @a)->flush; + +If C<Tie::File>'s memory limit is large enough, all the writing will +done in memory. Then, when you call C<-E<gt>flush>, the entire file +will be rewritten in a single pass. + +(Actually, the preceding discussion is something of a fib. You don't +need to enable deferred writing to get good performance for this +common case, because C<Tie::File> will do it for you automatically +unless you specifically tell it not to. See L<"autodeferring">, +below.) + +Calling C<-E<gt>flush> returns the array to immediate-write mode. If +you wish to discard the deferred writes, you may call C<-E<gt>discard> +instead of C<-E<gt>flush>. Note that in some cases, some of the data +will have been written already, and it will be too late for +C<-E<gt>discard> to discard all the changes. Support for +C<-E<gt>discard> may be withdrawn in a future version of C<Tie::File>. + +Deferred writes are cached in memory up to the limit specified by the +C<dw_size> option (see above). If the deferred-write buffer is full +and you try to write still more deferred data, the buffer will be +flushed. All buffered data will be written immediately, the buffer +will be emptied, and the now-empty space will be used for future +deferred writes. + +If the deferred-write buffer isn't yet full, but the total size of the +buffer and the read cache would exceed the C<memory> limit, the oldest +records will be expired from the read cache until the total size is +under the limit. + +C<push>, C<pop>, C<shift>, C<unshift>, and C<splice> cannot be +deferred. When you perform one of these operations, any deferred data +is written to the file and the operation is performed immediately. +This may change in a future version. + +If you resize the array with deferred writing enabled, the file will +be resized immediately, but deferred records will not be written. +This has a surprising consequence: C<@a = (...)> erases the file +immediately, but the writing of the actual data is deferred. This +might be a bug. If it is a bug, it will be fixed in a future version. + +=head2 Autodeferring + +C<Tie::File> tries to guess when deferred writing might be helpful, +and to turn it on and off automatically. + + for (@a) { + $_ = "> $_"; + } + +In this example, only the first two assignments will be done +immediately; after this, all the changes to the file will be deferred +up to the user-specified memory limit. + +You should usually be able to ignore this and just use the module +without thinking about deferring. However, special applications may +require fine control over which writes are deferred, or may require +that all writes be immediate. To disable the autodeferment feature, +use + + (tied @o)->autodefer(0); + +or + + tie @array, 'Tie::File', $file, autodefer => 0; + + +Similarly, C<-E<gt>autodefer(1)> re-enables autodeferment, and +C<-E<gt>autodefer()> recovers the current value of the autodefer setting. + + +=head1 CONCURRENT ACCESS TO FILES + +Caching and deferred writing are inappropriate if you want the same +file to be accessed simultaneously from more than one process. Other +optimizations performed internally by this module are also +incompatible with concurrent access. A future version of this module will +support a C<concurrent =E<gt> 1> option that enables safe concurrent access. + +Previous versions of this documentation suggested using C<memory +=E<gt> 0> for safe concurrent access. This was mistaken. Tie::File +will not support safe concurrent access before version 0.98. + +=head1 CAVEATS + +(That's Latin for 'warnings'.) + +=over 4 + +=item * + +Reasonable effort was made to make this module efficient. Nevertheless, +changing the size of a record in the middle of a large file will +always be fairly slow, because everything after the new record must be +moved. + +=item * + +The behavior of tied arrays is not precisely the same as for regular +arrays. For example: + + # This DOES print "How unusual!" + undef $a[10]; print "How unusual!\n" if defined $a[10]; + +C<undef>-ing a C<Tie::File> array element just blanks out the +corresponding record in the file. When you read it back again, you'll +get the empty string, so the supposedly-C<undef>'ed value will be +defined. Similarly, if you have C<autochomp> disabled, then + + # This DOES print "How unusual!" if 'autochomp' is disabled + undef $a[10]; + print "How unusual!\n" if $a[10]; + +Because when C<autochomp> is disabled, C<$a[10]> will read back as +C<"\n"> (or whatever the record separator string is.) + +There are other minor differences, particularly regarding C<exists> +and C<delete>, but in general, the correspondence is extremely close. + +=item * + +I have supposed that since this module is concerned with file I/O, +almost all normal use of it will be heavily I/O bound. This means +that the time to maintain complicated data structures inside the +module will be dominated by the time to actually perform the I/O. +When there was an opportunity to spend CPU time to avoid doing I/O, I +usually tried to take it. + +=item * + +You might be tempted to think that deferred writing is like +transactions, with C<flush> as C<commit> and C<discard> as +C<rollback>, but it isn't, so don't. + +=item * + +There is a large memory overhead for each record offset and for each +cache entry: about 310 bytes per cached data record, and about 21 bytes per offset table entry. + +The per-record overhead will limit the maximum number of records you +can access per file. Note that I<accessing> the length of the array +via C<$x = scalar @tied_file> accesses B<all> records and stores their +offsets. The same for C<foreach (@tied_file)>, even if you exit the +loop early. + +=back + +=head1 SUBCLASSING + +This version promises absolutely nothing about the internals, which +may change without notice. A future version of the module will have a +well-defined and stable subclassing API. + +=head1 WHAT ABOUT C<DB_File>? + +People sometimes point out that L<DB_File> will do something similar, +and ask why C<Tie::File> module is necessary. + +There are a number of reasons that you might prefer C<Tie::File>. +A list is available at C<http://perl.plover.com/TieFile/why-not-DB_File>. + +=head1 AUTHOR + +Mark Jason Dominus + +To contact the author, send email to: C<mjd-perl-tiefile+@plover.com> + +To receive an announcement whenever a new version of this module is +released, send a blank email message to +C<mjd-perl-tiefile-subscribe@plover.com>. + +The most recent version of this module, including documentation and +any news of importance, will be available at + + http://perl.plover.com/TieFile/ + + +=head1 LICENSE + +C<Tie::File> version 0.97 is copyright (C) 2003 Mark Jason Dominus. + +This library is free software; you may redistribute it and/or modify +it under the same terms as Perl itself. + +These terms are your choice of any of (1) the Perl Artistic Licence, +or (2) version 2 of the GNU General Public License as published by the +Free Software Foundation, or (3) any later version of the GNU General +Public License. + +This library is distributed in the hope that it will be useful, +but WITHOUT ANY WARRANTY; without even the implied warranty of +MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the +GNU General Public License for more details. + +You should have received a copy of the GNU General Public License +along with this library program; it should be in the file C<COPYING>. +If not, write to the Free Software Foundation, Inc., 51 Franklin Street, +Fifth Floor, Boston, MA 02110-1301, USA + +For licensing inquiries, contact the author at: + + Mark Jason Dominus + 255 S. Warnock St. + Philadelphia, PA 19107 + +=head1 WARRANTY + +C<Tie::File> version 0.97 comes with ABSOLUTELY NO WARRANTY. +For details, see the license. + +=head1 THANKS + +Gigantic thanks to Jarkko Hietaniemi, for agreeing to put this in the +core when I hadn't written it yet, and for generally being helpful, +supportive, and competent. (Usually the rule is "choose any one.") +Also big thanks to Abhijit Menon-Sen for all of the same things. + +Special thanks to Craig Berry and Peter Prymmer (for VMS portability +help), Randy Kobes (for Win32 portability help), Clinton Pierce and +Autrijus Tang (for heroic eleventh-hour Win32 testing above and beyond +the call of duty), Michael G Schwern (for testing advice), and the +rest of the CPAN testers (for testing generally). + +Special thanks to Tels for suggesting several speed and memory +optimizations. + +Additional thanks to: +Edward Avis / +Mattia Barbon / +Tom Christiansen / +Gerrit Haase / +Gurusamy Sarathy / +Jarkko Hietaniemi (again) / +Nikola Knezevic / +John Kominetz / +Nick Ing-Simmons / +Tassilo von Parseval / +H. Dieter Pearcey / +Slaven Rezic / +Eric Roode / +Peter Scott / +Peter Somu / +Autrijus Tang (again) / +Tels (again) / +Juerd Waalboer + +=head1 TODO + +More tests. (Stuff I didn't think of yet.) + +Paragraph mode? + +Fixed-length mode. Leave-blanks mode. + +Maybe an autolocking mode? + +For many common uses of the module, the read cache is a liability. +For example, a program that inserts a single record, or that scans the +file once, will have a cache hit rate of zero. This suggests a major +optimization: The cache should be initially disabled. Here's a hybrid +approach: Initially, the cache is disabled, but the cache code +maintains statistics about how high the hit rate would be *if* it were +enabled. When it sees the hit rate get high enough, it enables +itself. The STAT comments in this code are the beginning of an +implementation of this. + +Record locking with fcntl()? Then the module might support an undo +log and get real transactions. What a tour de force that would be. + +Keeping track of the highest cached record. This would allow reads-in-a-row +to skip the cache lookup faster (if reading from 1..N with empty cache at +start, the last cached value will be always N-1). + +More tests. + +=cut + diff --git a/Master/tlpkg/tlperl0/lib/Tie/Handle.pm b/Master/tlpkg/tlperl0/lib/Tie/Handle.pm new file mode 100755 index 00000000000..4e92f75a3d1 --- /dev/null +++ b/Master/tlpkg/tlperl0/lib/Tie/Handle.pm @@ -0,0 +1,201 @@ +package Tie::Handle; + +use 5.006_001; +our $VERSION = '4.2'; + +# Tie::StdHandle used to be inside Tie::Handle. For backwards compatibility +# loading Tie::Handle has to make Tie::StdHandle available. +use Tie::StdHandle; + +=head1 NAME + +Tie::Handle - base class definitions for tied handles + +=head1 SYNOPSIS + + package NewHandle; + require Tie::Handle; + + @ISA = qw(Tie::Handle); + + sub READ { ... } # Provide a needed method + sub TIEHANDLE { ... } # Overrides inherited method + + + package main; + + tie *FH, 'NewHandle'; + +=head1 DESCRIPTION + +This module provides some skeletal methods for handle-tying classes. See +L<perltie> for a list of the functions required in tying a handle to a package. +The basic B<Tie::Handle> package provides a C<new> method, as well as methods +C<TIEHANDLE>, C<PRINT>, C<PRINTF> and C<GETC>. + +For developers wishing to write their own tied-handle classes, the methods +are summarized below. The L<perltie> section not only documents these, but +has sample code as well: + +=over 4 + +=item TIEHANDLE classname, LIST + +The method invoked by the command C<tie *glob, classname>. Associates a new +glob instance with the specified class. C<LIST> would represent additional +arguments (along the lines of L<AnyDBM_File> and compatriots) needed to +complete the association. + +=item WRITE this, scalar, length, offset + +Write I<length> bytes of data from I<scalar> starting at I<offset>. + +=item PRINT this, LIST + +Print the values in I<LIST> + +=item PRINTF this, format, LIST + +Print the values in I<LIST> using I<format> + +=item READ this, scalar, length, offset + +Read I<length> bytes of data into I<scalar> starting at I<offset>. + +=item READLINE this + +Read a single line + +=item GETC this + +Get a single character + +=item CLOSE this + +Close the handle + +=item OPEN this, filename + +(Re-)open the handle + +=item BINMODE this + +Specify content is binary + +=item EOF this + +Test for end of file. + +=item TELL this + +Return position in the file. + +=item SEEK this, offset, whence + +Position the file. + +Test for end of file. + +=item DESTROY this + +Free the storage associated with the tied handle referenced by I<this>. +This is rarely needed, as Perl manages its memory quite well. But the +option exists, should a class wish to perform specific actions upon the +destruction of an instance. + +=back + +=head1 MORE INFORMATION + +The L<perltie> section contains an example of tying handles. + +=head1 COMPATIBILITY + +This version of Tie::Handle is neither related to nor compatible with +the Tie::Handle (3.0) module available on CPAN. It was due to an +accident that two modules with the same name appeared. The namespace +clash has been cleared in favor of this module that comes with the +perl core in September 2000 and accordingly the version number has +been bumped up to 4.0. + +=cut + +use Carp; +use warnings::register; + +sub new { + my $pkg = shift; + $pkg->TIEHANDLE(@_); +} + +# "Grandfather" the new, a la Tie::Hash + +sub TIEHANDLE { + my $pkg = shift; + if (defined &{"{$pkg}::new"}) { + warnings::warnif("WARNING: calling ${pkg}->new since ${pkg}->TIEHANDLE is missing"); + $pkg->new(@_); + } + else { + croak "$pkg doesn't define a TIEHANDLE method"; + } +} + +sub PRINT { + my $self = shift; + if($self->can('WRITE') != \&WRITE) { + my $buf = join(defined $, ? $, : "",@_); + $buf .= $\ if defined $\; + $self->WRITE($buf,length($buf),0); + } + else { + croak ref($self)," doesn't define a PRINT method"; + } +} + +sub PRINTF { + my $self = shift; + + if($self->can('WRITE') != \&WRITE) { + my $buf = sprintf(shift,@_); + $self->WRITE($buf,length($buf),0); + } + else { + croak ref($self)," doesn't define a PRINTF method"; + } +} + +sub READLINE { + my $pkg = ref $_[0]; + croak "$pkg doesn't define a READLINE method"; +} + +sub GETC { + my $self = shift; + + if($self->can('READ') != \&READ) { + my $buf; + $self->READ($buf,1); + return $buf; + } + else { + croak ref($self)," doesn't define a GETC method"; + } +} + +sub READ { + my $pkg = ref $_[0]; + croak "$pkg doesn't define a READ method"; +} + +sub WRITE { + my $pkg = ref $_[0]; + croak "$pkg doesn't define a WRITE method"; +} + +sub CLOSE { + my $pkg = ref $_[0]; + croak "$pkg doesn't define a CLOSE method"; +} + +1; diff --git a/Master/tlpkg/tlperl0/lib/Tie/Hash.pm b/Master/tlpkg/tlperl0/lib/Tie/Hash.pm new file mode 100755 index 00000000000..1ca8887e7e3 --- /dev/null +++ b/Master/tlpkg/tlperl0/lib/Tie/Hash.pm @@ -0,0 +1,257 @@ +package Tie::Hash; + +our $VERSION = '1.03'; + +=head1 NAME + +Tie::Hash, Tie::StdHash, Tie::ExtraHash - base class definitions for tied hashes + +=head1 SYNOPSIS + + package NewHash; + require Tie::Hash; + + @ISA = qw(Tie::Hash); + + sub DELETE { ... } # Provides needed method + sub CLEAR { ... } # Overrides inherited method + + + package NewStdHash; + require Tie::Hash; + + @ISA = qw(Tie::StdHash); + + # All methods provided by default, define only those needing overrides + # Accessors access the storage in %{$_[0]}; + # TIEHASH should return a reference to the actual storage + sub DELETE { ... } + + package NewExtraHash; + require Tie::Hash; + + @ISA = qw(Tie::ExtraHash); + + # All methods provided by default, define only those needing overrides + # Accessors access the storage in %{$_[0][0]}; + # TIEHASH should return an array reference with the first element being + # the reference to the actual storage + sub DELETE { + $_[0][1]->('del', $_[0][0], $_[1]); # Call the report writer + delete $_[0][0]->{$_[1]}; # $_[0]->SUPER::DELETE($_[1]) + } + + + package main; + + tie %new_hash, 'NewHash'; + tie %new_std_hash, 'NewStdHash'; + tie %new_extra_hash, 'NewExtraHash', + sub {warn "Doing \U$_[1]\E of $_[2].\n"}; + +=head1 DESCRIPTION + +This module provides some skeletal methods for hash-tying classes. See +L<perltie> for a list of the functions required in order to tie a hash +to a package. The basic B<Tie::Hash> package provides a C<new> method, as well +as methods C<TIEHASH>, C<EXISTS> and C<CLEAR>. The B<Tie::StdHash> and +B<Tie::ExtraHash> packages +provide most methods for hashes described in L<perltie> (the exceptions +are C<UNTIE> and C<DESTROY>). They cause tied hashes to behave exactly like standard hashes, +and allow for selective overwriting of methods. B<Tie::Hash> grandfathers the +C<new> method: it is used if C<TIEHASH> is not defined +in the case a class forgets to include a C<TIEHASH> method. + +For developers wishing to write their own tied hashes, the required methods +are briefly defined below. See the L<perltie> section for more detailed +descriptive, as well as example code: + +=over 4 + +=item TIEHASH classname, LIST + +The method invoked by the command C<tie %hash, classname>. Associates a new +hash instance with the specified class. C<LIST> would represent additional +arguments (along the lines of L<AnyDBM_File> and compatriots) needed to +complete the association. + +=item STORE this, key, value + +Store datum I<value> into I<key> for the tied hash I<this>. + +=item FETCH this, key + +Retrieve the datum in I<key> for the tied hash I<this>. + +=item FIRSTKEY this + +Return the first key in the hash. + +=item NEXTKEY this, lastkey + +Return the next key in the hash. + +=item EXISTS this, key + +Verify that I<key> exists with the tied hash I<this>. + +The B<Tie::Hash> implementation is a stub that simply croaks. + +=item DELETE this, key + +Delete the key I<key> from the tied hash I<this>. + +=item CLEAR this + +Clear all values from the tied hash I<this>. + +=item SCALAR this + +Returns what evaluating the hash in scalar context yields. + +B<Tie::Hash> does not implement this method (but B<Tie::StdHash> +and B<Tie::ExtraHash> do). + +=back + +=head1 Inheriting from B<Tie::StdHash> + +The accessor methods assume that the actual storage for the data in the tied +hash is in the hash referenced by C<tied(%tiedhash)>. Thus overwritten +C<TIEHASH> method should return a hash reference, and the remaining methods +should operate on the hash referenced by the first argument: + + package ReportHash; + our @ISA = 'Tie::StdHash'; + + sub TIEHASH { + my $storage = bless {}, shift; + warn "New ReportHash created, stored in $storage.\n"; + $storage + } + sub STORE { + warn "Storing data with key $_[1] at $_[0].\n"; + $_[0]{$_[1]} = $_[2] + } + + +=head1 Inheriting from B<Tie::ExtraHash> + +The accessor methods assume that the actual storage for the data in the tied +hash is in the hash referenced by C<(tied(%tiedhash))-E<gt>[0]>. Thus overwritten +C<TIEHASH> method should return an array reference with the first +element being a hash reference, and the remaining methods should operate on the +hash C<< %{ $_[0]->[0] } >>: + + package ReportHash; + our @ISA = 'Tie::ExtraHash'; + + sub TIEHASH { + my $class = shift; + my $storage = bless [{}, @_], $class; + warn "New ReportHash created, stored in $storage.\n"; + $storage; + } + sub STORE { + warn "Storing data with key $_[1] at $_[0].\n"; + $_[0][0]{$_[1]} = $_[2] + } + +The default C<TIEHASH> method stores "extra" arguments to tie() starting +from offset 1 in the array referenced by C<tied(%tiedhash)>; this is the +same storage algorithm as in TIEHASH subroutine above. Hence, a typical +package inheriting from B<Tie::ExtraHash> does not need to overwrite this +method. + +=head1 C<SCALAR>, C<UNTIE> and C<DESTROY> + +The methods C<UNTIE> and C<DESTROY> are not defined in B<Tie::Hash>, +B<Tie::StdHash>, or B<Tie::ExtraHash>. Tied hashes do not require +presence of these methods, but if defined, the methods will be called in +proper time, see L<perltie>. + +C<SCALAR> is only defined in B<Tie::StdHash> and B<Tie::ExtraHash>. + +If needed, these methods should be defined by the package inheriting from +B<Tie::Hash>, B<Tie::StdHash>, or B<Tie::ExtraHash>. See L<perltie/"SCALAR"> +to find out what happens when C<SCALAR> does not exist. + +=head1 MORE INFORMATION + +The packages relating to various DBM-related implementations (F<DB_File>, +F<NDBM_File>, etc.) show examples of general tied hashes, as does the +L<Config> module. While these do not utilize B<Tie::Hash>, they serve as +good working examples. + +=cut + +use Carp; +use warnings::register; + +sub new { + my $pkg = shift; + $pkg->TIEHASH(@_); +} + +# Grandfather "new" + +sub TIEHASH { + my $pkg = shift; + if (defined &{"${pkg}::new"}) { + warnings::warnif("WARNING: calling ${pkg}->new since ${pkg}->TIEHASH is missing"); + $pkg->new(@_); + } + else { + croak "$pkg doesn't define a TIEHASH method"; + } +} + +sub EXISTS { + my $pkg = ref $_[0]; + croak "$pkg doesn't define an EXISTS method"; +} + +sub CLEAR { + my $self = shift; + my $key = $self->FIRSTKEY(@_); + my @keys; + + while (defined $key) { + push @keys, $key; + $key = $self->NEXTKEY(@_, $key); + } + foreach $key (@keys) { + $self->DELETE(@_, $key); + } +} + +# The Tie::StdHash package implements standard perl hash behaviour. +# It exists to act as a base class for classes which only wish to +# alter some parts of their behaviour. + +package Tie::StdHash; +# @ISA = qw(Tie::Hash); # would inherit new() only + +sub TIEHASH { bless {}, $_[0] } +sub STORE { $_[0]->{$_[1]} = $_[2] } +sub FETCH { $_[0]->{$_[1]} } +sub FIRSTKEY { my $a = scalar keys %{$_[0]}; each %{$_[0]} } +sub NEXTKEY { each %{$_[0]} } +sub EXISTS { exists $_[0]->{$_[1]} } +sub DELETE { delete $_[0]->{$_[1]} } +sub CLEAR { %{$_[0]} = () } +sub SCALAR { scalar %{$_[0]} } + +package Tie::ExtraHash; + +sub TIEHASH { my $p = shift; bless [{}, @_], $p } +sub STORE { $_[0][0]{$_[1]} = $_[2] } +sub FETCH { $_[0][0]{$_[1]} } +sub FIRSTKEY { my $a = scalar keys %{$_[0][0]}; each %{$_[0][0]} } +sub NEXTKEY { each %{$_[0][0]} } +sub EXISTS { exists $_[0][0]->{$_[1]} } +sub DELETE { delete $_[0][0]->{$_[1]} } +sub CLEAR { %{$_[0][0]} = () } +sub SCALAR { scalar %{$_[0][0]} } + +1; diff --git a/Master/tlpkg/tlperl0/lib/Tie/Hash/NamedCapture.pm b/Master/tlpkg/tlperl0/lib/Tie/Hash/NamedCapture.pm new file mode 100755 index 00000000000..58ae743d874 --- /dev/null +++ b/Master/tlpkg/tlperl0/lib/Tie/Hash/NamedCapture.pm @@ -0,0 +1,62 @@ +package Tie::Hash::NamedCapture; + +our $VERSION = "0.06"; + +# The real meat implemented in XS in universal.c in the core, but this +# method was left behind because gv.c expects a Purl-Perl method in +# this package when it loads the tie magic for %+ and %- + +my ($one, $all) = Tie::Hash::NamedCapture::flags(); + +sub TIEHASH { + my ($pkg, %arg) = @_; + my $flag = $arg{all} ? $all : $one; + bless \$flag => $pkg; +} + +tie %+, __PACKAGE__; +tie %-, __PACKAGE__, all => 1; + +1; + +__END__ + +=head1 NAME + +Tie::Hash::NamedCapture - Named regexp capture buffers + +=head1 SYNOPSIS + + tie my %hash, "Tie::Hash::NamedCapture"; + # %hash now behaves like %+ + + tie my %hash, "Tie::Hash::NamedCapture", all => 1; + # %hash now access buffers from regexp in $qr like %- + +=head1 DESCRIPTION + +This module is used to implement the special hashes C<%+> and C<%->, but it +can be used to tie other variables as you choose. + +When the C<all> parameter is provided, then the tied hash elements will be +array refs listing the contents of each capture buffer whose name is the +same as the associated hash key. If none of these buffers were involved in +the match, the contents of that array ref will be as many C<undef> values +as there are capture buffers with that name. In other words, the tied hash +will behave as C<%->. + +When the C<all> parameter is omitted or false, then the tied hash elements +will be the contents of the leftmost defined buffer with the name of the +associated hash key. In other words, the tied hash will behave as +C<%+>. + +The keys of C<%->-like hashes correspond to all buffer names found in the +regular expression; the keys of C<%+>-like hashes list only the names of +buffers that have captured (and that are thus associated to defined values). + +=head1 SEE ALSO + +L<perlreapi>, L<re>, L<perlmodlib/Pragmatic Modules>, L<perlvar/"%+">, +L<perlvar/"%-">. + +=cut diff --git a/Master/tlpkg/tlperl0/lib/Tie/Memoize.pm b/Master/tlpkg/tlperl0/lib/Tie/Memoize.pm new file mode 100755 index 00000000000..dbe14428280 --- /dev/null +++ b/Master/tlpkg/tlperl0/lib/Tie/Memoize.pm @@ -0,0 +1,128 @@ +use strict; +package Tie::Memoize; +use Tie::Hash; +our @ISA = 'Tie::ExtraHash'; +our $VERSION = '1.1'; + +our $exists_token = \undef; + +sub croak {require Carp; goto &Carp::croak} + +# Format: [0: STORAGE, 1: EXISTS-CACHE, 2: FETCH_function; +# 3: EXISTS_function, 4: DATA, 5: EXISTS_different ] + +sub FETCH { + my ($h,$key) = ($_[0][0], $_[1]); + my $res = $h->{$key}; + return $res if defined $res; # Shortcut if accessible + return $res if exists $h->{$key}; # Accessible, but undef + my $cache = $_[0][1]{$key}; + return if defined $cache and not $cache; # Known to not exist + my @res = $_[0][2]->($key, $_[0][4]); # Autoload + $_[0][1]{$key} = 0, return unless @res; # Cache non-existence + delete $_[0][1]{$key}; # Clear existence cache, not needed any more + $_[0][0]{$key} = $res[0]; # Store data and return +} + +sub EXISTS { + my ($a,$key) = (shift, shift); + return 1 if exists $a->[0]{$key}; # Have data + my $cache = $a->[1]{$key}; + return $cache if defined $cache; # Existence cache + my @res = $a->[3]($key,$a->[4]); + $a->[1]{$key} = 0, return unless @res; # Cache non-existence + # Now we know it exists + return ($a->[1]{$key} = 1) if $a->[5]; # Only existence reported + # Now know the value + $a->[0]{$key} = $res[0]; # Store data + return 1 +} + +sub TIEHASH { + croak 'syntax: tie %hash, \'Tie::AutoLoad\', \&fetch_subr' if @_ < 2; + croak 'syntax: tie %hash, \'Tie::AutoLoad\', \&fetch_subr, $data, \&exists_subr, \%data_cache, \%existence_cache' if @_ > 6; + push @_, undef if @_ < 3; # Data + push @_, $_[1] if @_ < 4; # exists + push @_, {} while @_ < 6; # initial value and caches + bless [ @_[4,5,1,3,2], $_[1] ne $_[3]], $_[0] +} + +1; + +=head1 NAME + +Tie::Memoize - add data to hash when needed + +=head1 SYNOPSIS + + require Tie::Memoize; + tie %hash, 'Tie::Memoize', + \&fetch, # The rest is optional + $DATA, \&exists, + {%ini_value}, {%ini_existence}; + +=head1 DESCRIPTION + +This package allows a tied hash to autoload its values on the first access, +and to use the cached value on the following accesses. + +Only read-accesses (via fetching the value or C<exists>) result in calls to +the functions; the modify-accesses are performed as on a normal hash. + +The required arguments during C<tie> are the hash, the package, and +the reference to the C<FETCH>ing function. The optional arguments are +an arbitrary scalar $data, the reference to the C<EXISTS> function, +and initial values of the hash and of the existence cache. + +Both the C<FETCH>ing function and the C<EXISTS> functions have the +same signature: the arguments are C<$key, $data>; $data is the same +value as given as argument during tie()ing. Both functions should +return an empty list if the value does not exist. If C<EXISTS> +function is different from the C<FETCH>ing function, it should return +a TRUE value on success. The C<FETCH>ing function should return the +intended value if the key is valid. + +=head1 Inheriting from B<Tie::Memoize> + +The structure of the tied() data is an array reference with elements + + 0: cache of known values + 1: cache of known existence of keys + 2: FETCH function + 3: EXISTS function + 4: $data + +The rest is for internal usage of this package. In particular, if +TIEHASH is overwritten, it should call SUPER::TIEHASH. + +=head1 EXAMPLE + + sub slurp { + my ($key, $dir) = shift; + open my $h, '<', "$dir/$key" or return; + local $/; <$h> # slurp it all + } + sub exists { my ($key, $dir) = shift; return -f "$dir/$key" } + + tie %hash, 'Tie::Memoize', \&slurp, $directory, \&exists, + { fake_file1 => $content1, fake_file2 => $content2 }, + { pretend_does_not_exists => 0, known_to_exist => 1 }; + +This example treats the slightly modified contents of $directory as a +hash. The modifications are that the keys F<fake_file1> and +F<fake_file2> fetch values $content1 and $content2, and +F<pretend_does_not_exists> will never be accessed. Additionally, the +existence of F<known_to_exist> is never checked (so if it does not +exists when its content is needed, the user of %hash may be confused). + +=head1 BUGS + +FIRSTKEY and NEXTKEY methods go through the keys which were already read, +not all the possible keys of the hash. + +=head1 AUTHOR + +Ilya Zakharevich L<mailto:perl-module-hash-memoize@ilyaz.org>. + +=cut + diff --git a/Master/tlpkg/tlperl0/lib/Tie/RefHash.pm b/Master/tlpkg/tlperl0/lib/Tie/RefHash.pm new file mode 100755 index 00000000000..f95bf41efdf --- /dev/null +++ b/Master/tlpkg/tlperl0/lib/Tie/RefHash.pm @@ -0,0 +1,274 @@ +package Tie::RefHash; + +use vars qw/$VERSION/; + +$VERSION = "1.38"; + +use 5.005; + +=head1 NAME + +Tie::RefHash - use references as hash keys + +=head1 SYNOPSIS + + require 5.004; + use Tie::RefHash; + tie HASHVARIABLE, 'Tie::RefHash', LIST; + tie HASHVARIABLE, 'Tie::RefHash::Nestable', LIST; + + untie HASHVARIABLE; + +=head1 DESCRIPTION + +This module provides the ability to use references as hash keys if you +first C<tie> the hash variable to this module. Normally, only the +keys of the tied hash itself are preserved as references; to use +references as keys in hashes-of-hashes, use Tie::RefHash::Nestable, +included as part of Tie::RefHash. + +It is implemented using the standard perl TIEHASH interface. Please +see the C<tie> entry in perlfunc(1) and perltie(1) for more information. + +The Nestable version works by looking for hash references being stored +and converting them to tied hashes so that they too can have +references as keys. This will happen without warning whenever you +store a reference to one of your own hashes in the tied hash. + +=head1 EXAMPLE + + use Tie::RefHash; + tie %h, 'Tie::RefHash'; + $a = []; + $b = {}; + $c = \*main; + $d = \"gunk"; + $e = sub { 'foo' }; + %h = ($a => 1, $b => 2, $c => 3, $d => 4, $e => 5); + $a->[0] = 'foo'; + $b->{foo} = 'bar'; + for (keys %h) { + print ref($_), "\n"; + } + + tie %h, 'Tie::RefHash::Nestable'; + $h{$a}->{$b} = 1; + for (keys %h, keys %{$h{$a}}) { + print ref($_), "\n"; + } + +=head1 THREAD SUPPORT + +L<Tie::RefHash> fully supports threading using the C<CLONE> method. + +=head1 STORABLE SUPPORT + +L<Storable> hooks are provided for semantically correct serialization and +cloning of tied refhashes. + +=head1 RELIC SUPPORT + +This version of Tie::RefHash seems to no longer work with 5.004. This has not +been throughly investigated. Patches welcome ;-) + +=head1 MAINTAINER + +Yuval Kogman E<lt>nothingmuch@woobling.orgE<gt> + +=head1 AUTHOR + +Gurusamy Sarathy gsar@activestate.com + +'Nestable' by Ed Avis ed@membled.com + +=head1 SEE ALSO + +perl(1), perlfunc(1), perltie(1) + +=cut + +use Tie::Hash; +use vars '@ISA'; +@ISA = qw(Tie::Hash); +use strict; +use Carp qw/croak/; + +BEGIN { + local $@; + # determine whether we need to take care of threads + use Config (); + my $usethreads = $Config::Config{usethreads}; # && exists $INC{"threads.pm"} + *_HAS_THREADS = $usethreads ? sub () { 1 } : sub () { 0 }; + *_HAS_SCALAR_UTIL = eval { require Scalar::Util; 1 } ? sub () { 1 } : sub () { 0 }; + *_HAS_WEAKEN = defined(&Scalar::Util::weaken) ? sub () { 1 } : sub () { 0 }; +} + +BEGIN { + # create a refaddr function + + local $@; + + if ( _HAS_SCALAR_UTIL ) { + Scalar::Util->import("refaddr"); + } else { + require overload; + + *refaddr = sub { + if ( overload::StrVal($_[0]) =~ /\( 0x ([a-zA-Z0-9]+) \)$/x) { + return $1; + } else { + die "couldn't parse StrVal: " . overload::StrVal($_[0]); + } + }; + } +} + +my (@thread_object_registry, $count); # used by the CLONE method to rehash the keys after their refaddr changed + +sub TIEHASH { + my $c = shift; + my $s = []; + bless $s, $c; + while (@_) { + $s->STORE(shift, shift); + } + + if (_HAS_THREADS ) { + + if ( _HAS_WEAKEN ) { + # remember the object so that we can rekey it on CLONE + push @thread_object_registry, $s; + # but make this a weak reference, so that there are no leaks + Scalar::Util::weaken( $thread_object_registry[-1] ); + + if ( ++$count > 1000 ) { + # this ensures we don't fill up with a huge array dead weakrefs + @thread_object_registry = grep { defined } @thread_object_registry; + $count = 0; + } + } else { + $count++; # used in the warning + } + } + + return $s; +} + +my $storable_format_version = join("/", __PACKAGE__, "0.01"); + +sub STORABLE_freeze { + my ( $self, $is_cloning ) = @_; + my ( $refs, $reg ) = @$self; + return ( $storable_format_version, [ values %$refs ], $reg ); +} + +sub STORABLE_thaw { + my ( $self, $is_cloning, $version, $refs, $reg ) = @_; + croak "incompatible versions of Tie::RefHash between freeze and thaw" + unless $version eq $storable_format_version; + + @$self = ( {}, $reg ); + $self->_reindex_keys( $refs ); +} + +sub CLONE { + my $pkg = shift; + + if ( $count and not _HAS_WEAKEN ) { + warn "Tie::RefHash is not threadsafe without Scalar::Util::weaken"; + } + + # when the thread has been cloned all the objects need to be updated. + # dead weakrefs are undefined, so we filter them out + @thread_object_registry = grep { defined && do { $_->_reindex_keys; 1 } } @thread_object_registry; + $count = 0; # we just cleaned up +} + +sub _reindex_keys { + my ( $self, $extra_keys ) = @_; + # rehash all the ref keys based on their new StrVal + %{ $self->[0] } = map { refaddr($_->[0]) => $_ } (values(%{ $self->[0] }), @{ $extra_keys || [] }); +} + +sub FETCH { + my($s, $k) = @_; + if (ref $k) { + my $kstr = refaddr($k); + if (defined $s->[0]{$kstr}) { + $s->[0]{$kstr}[1]; + } + else { + undef; + } + } + else { + $s->[1]{$k}; + } +} + +sub STORE { + my($s, $k, $v) = @_; + if (ref $k) { + $s->[0]{refaddr($k)} = [$k, $v]; + } + else { + $s->[1]{$k} = $v; + } + $v; +} + +sub DELETE { + my($s, $k) = @_; + (ref $k) + ? (delete($s->[0]{refaddr($k)}) || [])->[1] + : delete($s->[1]{$k}); +} + +sub EXISTS { + my($s, $k) = @_; + (ref $k) ? exists($s->[0]{refaddr($k)}) : exists($s->[1]{$k}); +} + +sub FIRSTKEY { + my $s = shift; + keys %{$s->[0]}; # reset iterator + keys %{$s->[1]}; # reset iterator + $s->[2] = 0; # flag for iteration, see NEXTKEY + $s->NEXTKEY; +} + +sub NEXTKEY { + my $s = shift; + my ($k, $v); + if (!$s->[2]) { + if (($k, $v) = each %{$s->[0]}) { + return $v->[0]; + } + else { + $s->[2] = 1; + } + } + return each %{$s->[1]}; +} + +sub CLEAR { + my $s = shift; + $s->[2] = 0; + %{$s->[0]} = (); + %{$s->[1]} = (); +} + +package Tie::RefHash::Nestable; +use vars '@ISA'; +@ISA = 'Tie::RefHash'; + +sub STORE { + my($s, $k, $v) = @_; + if (ref($v) eq 'HASH' and not tied %$v) { + my @elems = %$v; + tie %$v, ref($s), @elems; + } + $s->SUPER::STORE($k, $v); +} + +1; diff --git a/Master/tlpkg/tlperl0/lib/Tie/Registry.pm b/Master/tlpkg/tlperl0/lib/Tie/Registry.pm new file mode 100755 index 00000000000..2ded338d342 --- /dev/null +++ b/Master/tlpkg/tlperl0/lib/Tie/Registry.pm @@ -0,0 +1,45 @@ +package Tie::Registry; + +# Tie/Registry.pm -- Provides backward compatibility for Win32::TieRegistry +# that was called Tie::Registry prior to version 0.20. +# by Tye McQueen, tye@metronet.com, see http://www.metronet.com/~tye/. + +use strict; +use Carp; + +use vars qw( $VERSION @ISA ); +BEGIN { + require Win32::TieRegistry; + $VERSION = '0.15'; + @ISA = qw{Win32::TieRegistry}; +} + +sub import { + my $pkg = shift; + Win32::TieRegistry->import( ExportLevel => 1, SplitMultis => 0, @_ ); +} + +1; + +__END__ + +=pod + +=head1 NAME + +Tie::Registry - Legacy interface to Win32::TieRegistry (DEPRECATED) + +=head1 DESCRIPTION + +This module provides backward compatibility for L<Win32::TieRegistry> +that was called Tie::Registry prior to version 0.20. + +=head1 AUTHOR + +Tye McQueen E<lt>tye@metronet.comE<gt> + +=head1 COPYRIGHT + +Copyright 1999 Tye McQueen. + +=cut diff --git a/Master/tlpkg/tlperl0/lib/Tie/Scalar.pm b/Master/tlpkg/tlperl0/lib/Tie/Scalar.pm new file mode 100755 index 00000000000..9bf03f9105c --- /dev/null +++ b/Master/tlpkg/tlperl0/lib/Tie/Scalar.pm @@ -0,0 +1,140 @@ +package Tie::Scalar; + +our $VERSION = '1.01'; + +=head1 NAME + +Tie::Scalar, Tie::StdScalar - base class definitions for tied scalars + +=head1 SYNOPSIS + + package NewScalar; + require Tie::Scalar; + + @ISA = qw(Tie::Scalar); + + sub FETCH { ... } # Provide a needed method + sub TIESCALAR { ... } # Overrides inherited method + + + package NewStdScalar; + require Tie::Scalar; + + @ISA = qw(Tie::StdScalar); + + # All methods provided by default, so define only what needs be overridden + sub FETCH { ... } + + + package main; + + tie $new_scalar, 'NewScalar'; + tie $new_std_scalar, 'NewStdScalar'; + +=head1 DESCRIPTION + +This module provides some skeletal methods for scalar-tying classes. See +L<perltie> for a list of the functions required in tying a scalar to a +package. The basic B<Tie::Scalar> package provides a C<new> method, as well +as methods C<TIESCALAR>, C<FETCH> and C<STORE>. The B<Tie::StdScalar> +package provides all the methods specified in L<perltie>. It inherits from +B<Tie::Scalar> and causes scalars tied to it to behave exactly like the +built-in scalars, allowing for selective overloading of methods. The C<new> +method is provided as a means of grandfathering, for classes that forget to +provide their own C<TIESCALAR> method. + +For developers wishing to write their own tied-scalar classes, the methods +are summarized below. The L<perltie> section not only documents these, but +has sample code as well: + +=over 4 + +=item TIESCALAR classname, LIST + +The method invoked by the command C<tie $scalar, classname>. Associates a new +scalar instance with the specified class. C<LIST> would represent additional +arguments (along the lines of L<AnyDBM_File> and compatriots) needed to +complete the association. + +=item FETCH this + +Retrieve the value of the tied scalar referenced by I<this>. + +=item STORE this, value + +Store data I<value> in the tied scalar referenced by I<this>. + +=item DESTROY this + +Free the storage associated with the tied scalar referenced by I<this>. +This is rarely needed, as Perl manages its memory quite well. But the +option exists, should a class wish to perform specific actions upon the +destruction of an instance. + +=back + +=head1 MORE INFORMATION + +The L<perltie> section uses a good example of tying scalars by associating +process IDs with priority. + +=cut + +use Carp; +use warnings::register; + +sub new { + my $pkg = shift; + $pkg->TIESCALAR(@_); +} + +# "Grandfather" the new, a la Tie::Hash + +sub TIESCALAR { + my $pkg = shift; + if ($pkg->can('new') and $pkg ne __PACKAGE__) { + warnings::warnif("WARNING: calling ${pkg}->new since ${pkg}->TIESCALAR is missing"); + $pkg->new(@_); + } + else { + croak "$pkg doesn't define a TIESCALAR method"; + } +} + +sub FETCH { + my $pkg = ref $_[0]; + croak "$pkg doesn't define a FETCH method"; +} + +sub STORE { + my $pkg = ref $_[0]; + croak "$pkg doesn't define a STORE method"; +} + +# +# The Tie::StdScalar package provides scalars that behave exactly like +# Perl's built-in scalars. Good base to inherit from, if you're only going to +# tweak a small bit. +# +package Tie::StdScalar; +@ISA = qw(Tie::Scalar); + +sub TIESCALAR { + my $class = shift; + my $instance = shift || undef; + return bless \$instance => $class; +} + +sub FETCH { + return ${$_[0]}; +} + +sub STORE { + ${$_[0]} = $_[1]; +} + +sub DESTROY { + undef ${$_[0]}; +} + +1; diff --git a/Master/tlpkg/tlperl0/lib/Tie/StdHandle.pm b/Master/tlpkg/tlperl0/lib/Tie/StdHandle.pm new file mode 100755 index 00000000000..3a1a3db4788 --- /dev/null +++ b/Master/tlpkg/tlperl0/lib/Tie/StdHandle.pm @@ -0,0 +1,71 @@ +package Tie::StdHandle; + +use strict; + +use Tie::Handle; +use vars qw(@ISA $VERSION); +@ISA = 'Tie::Handle'; +$VERSION = '4.2'; + +=head1 NAME + +Tie::StdHandle - base class definitions for tied handles + +=head1 SYNOPSIS + + package NewHandle; + require Tie::Handle; + + @ISA = qw(Tie::Handle); + + sub READ { ... } # Provide a needed method + sub TIEHANDLE { ... } # Overrides inherited method + + + package main; + + tie *FH, 'NewHandle'; + +=head1 DESCRIPTION + +The B<Tie::StdHandle> package provide most methods for file handles described +in L<perltie> (the exceptions are C<UNTIE> and C<DESTROY>). It causes tied +file handles to behave exactly like standard file handles and allow for +selective overwriting of methods. + +=cut + +sub TIEHANDLE +{ + my $class = shift; + my $fh = \do { local *HANDLE}; + bless $fh,$class; + $fh->OPEN(@_) if (@_); + return $fh; +} + +sub EOF { eof($_[0]) } +sub TELL { tell($_[0]) } +sub FILENO { fileno($_[0]) } +sub SEEK { seek($_[0],$_[1],$_[2]) } +sub CLOSE { close($_[0]) } +sub BINMODE { binmode($_[0]) } + +sub OPEN +{ + $_[0]->CLOSE if defined($_[0]->FILENO); + @_ == 2 ? open($_[0], $_[1]) : open($_[0], $_[1], $_[2]); +} + +sub READ { read($_[0],$_[1],$_[2]) } +sub READLINE { my $fh = $_[0]; <$fh> } +sub GETC { getc($_[0]) } + +sub WRITE +{ + my $fh = $_[0]; + print $fh substr($_[1],0,$_[2]) +} + + +1; diff --git a/Master/tlpkg/tlperl0/lib/Tie/SubstrHash.pm b/Master/tlpkg/tlperl0/lib/Tie/SubstrHash.pm new file mode 100755 index 00000000000..476dd686787 --- /dev/null +++ b/Master/tlpkg/tlperl0/lib/Tie/SubstrHash.pm @@ -0,0 +1,215 @@ +package Tie::SubstrHash; + +our $VERSION = '1.00'; + +=head1 NAME + +Tie::SubstrHash - Fixed-table-size, fixed-key-length hashing + +=head1 SYNOPSIS + + require Tie::SubstrHash; + + tie %myhash, 'Tie::SubstrHash', $key_len, $value_len, $table_size; + +=head1 DESCRIPTION + +The B<Tie::SubstrHash> package provides a hash-table-like interface to +an array of determinate size, with constant key size and record size. + +Upon tying a new hash to this package, the developer must specify the +size of the keys that will be used, the size of the value fields that the +keys will index, and the size of the overall table (in terms of key-value +pairs, not size in hard memory). I<These values will not change for the +duration of the tied hash>. The newly-allocated hash table may now have +data stored and retrieved. Efforts to store more than C<$table_size> +elements will result in a fatal error, as will efforts to store a value +not exactly C<$value_len> characters in length, or reference through a +key not exactly C<$key_len> characters in length. While these constraints +may seem excessive, the result is a hash table using much less internal +memory than an equivalent freely-allocated hash table. + +=head1 CAVEATS + +Because the current implementation uses the table and key sizes for the +hashing algorithm, there is no means by which to dynamically change the +value of any of the initialization parameters. + +The hash does not support exists(). + +=cut + +use Carp; + +sub TIEHASH { + my $pack = shift; + my ($klen, $vlen, $tsize) = @_; + my $rlen = 1 + $klen + $vlen; + $tsize = [$tsize, + findgteprime($tsize * 1.1)]; # Allow 10% empty. + local $self = bless ["\0", $klen, $vlen, $tsize, $rlen, 0, -1]; + $$self[0] x= $rlen * $tsize->[1]; + $self; +} + +sub CLEAR { + local($self) = @_; + $$self[0] = "\0" x ($$self[4] * $$self[3]->[1]); + $$self[5] = 0; + $$self[6] = -1; +} + +sub FETCH { + local($self,$key) = @_; + local($klen, $vlen, $tsize, $rlen) = @$self[1..4]; + &hashkey; + for (;;) { + $offset = $hash * $rlen; + $record = substr($$self[0], $offset, $rlen); + if (ord($record) == 0) { + return undef; + } + elsif (ord($record) == 1) { + } + elsif (substr($record, 1, $klen) eq $key) { + return substr($record, 1+$klen, $vlen); + } + &rehash; + } +} + +sub STORE { + local($self,$key,$val) = @_; + local($klen, $vlen, $tsize, $rlen) = @$self[1..4]; + croak("Table is full ($tsize->[0] elements)") if $$self[5] > $tsize->[0]; + croak(qq/Value "$val" is not $vlen characters long/) + if length($val) != $vlen; + my $writeoffset; + + &hashkey; + for (;;) { + $offset = $hash * $rlen; + $record = substr($$self[0], $offset, $rlen); + if (ord($record) == 0) { + $record = "\2". $key . $val; + die "panic" unless length($record) == $rlen; + $writeoffset = $offset unless defined $writeoffset; + substr($$self[0], $writeoffset, $rlen) = $record; + ++$$self[5]; + return; + } + elsif (ord($record) == 1) { + $writeoffset = $offset unless defined $writeoffset; + } + elsif (substr($record, 1, $klen) eq $key) { + $record = "\2". $key . $val; + die "panic" unless length($record) == $rlen; + substr($$self[0], $offset, $rlen) = $record; + return; + } + &rehash; + } +} + +sub DELETE { + local($self,$key) = @_; + local($klen, $vlen, $tsize, $rlen) = @$self[1..4]; + &hashkey; + for (;;) { + $offset = $hash * $rlen; + $record = substr($$self[0], $offset, $rlen); + if (ord($record) == 0) { + return undef; + } + elsif (ord($record) == 1) { + } + elsif (substr($record, 1, $klen) eq $key) { + substr($$self[0], $offset, 1) = "\1"; + return substr($record, 1+$klen, $vlen); + --$$self[5]; + } + &rehash; + } +} + +sub FIRSTKEY { + local($self) = @_; + $$self[6] = -1; + &NEXTKEY; +} + +sub NEXTKEY { + local($self) = @_; + local($klen, $vlen, $tsize, $rlen, $entries, $iterix) = @$self[1..6]; + for (++$iterix; $iterix < $tsize->[1]; ++$iterix) { + next unless substr($$self[0], $iterix * $rlen, 1) eq "\2"; + $$self[6] = $iterix; + return substr($$self[0], $iterix * $rlen + 1, $klen); + } + $$self[6] = -1; + undef; +} + +sub EXISTS { + croak "Tie::SubstrHash does not support exists()"; +} + +sub hashkey { + croak(qq/Key "$key" is not $klen characters long/) + if length($key) != $klen; + $hash = 2; + for (unpack('C*', $key)) { + $hash = $hash * 33 + $_; + &_hashwrap if $hash >= 1e13; + } + &_hashwrap if $hash >= $tsize->[1]; + $hash = 1 unless $hash; + $hashbase = $hash; +} + +sub _hashwrap { + $hash -= int($hash / $tsize->[1]) * $tsize->[1]; +} + +sub rehash { + $hash += $hashbase; + $hash -= $tsize->[1] if $hash >= $tsize->[1]; +} + +# using POSIX::ceil() would be too heavy, and not all platforms have it. +sub ceil { + my $num = shift; + $num = int($num + 1) unless $num == int $num; + return $num; +} + +# See: +# +# http://www-groups.dcs.st-andrews.ac.uk/~history/HistTopics/Prime_numbers.html +# + +sub findgteprime { # find the smallest prime integer greater than or equal to + use integer; + + my $num = ceil(shift); + return 2 if $num <= 2; + + $num++ unless $num % 2; + my $i; + my $sqrtnum = int sqrt $num; + my $sqrtnumsquared = $sqrtnum * $sqrtnum; + + NUM: + for (;; $num += 2) { + if ($sqrtnumsquared < $num) { + $sqrtnum++; + $sqrtnumsquared = $sqrtnum * $sqrtnum; + } + for ($i = 3; $i <= $sqrtnum; $i += 2) { + next NUM unless $num % $i; + } + return $num; + } +} + +1; diff --git a/Master/tlpkg/tlperl0/lib/Tie/Watch.pm b/Master/tlpkg/tlperl0/lib/Tie/Watch.pm new file mode 100755 index 00000000000..48f46acd2b0 --- /dev/null +++ b/Master/tlpkg/tlperl0/lib/Tie/Watch.pm @@ -0,0 +1,560 @@ +$Tie::Watch::VERSION = '1.2'; + +package Tie::Watch; + +=head1 NAME + + Tie::Watch - place watchpoints on Perl variables. + +=head1 SYNOPSIS + + use Tie::Watch; + + $watch = Tie::Watch->new( + -variable => \$frog, + -debug => 1, + -shadow => 0, + -fetch => [\&fetch, 'arg1', 'arg2', ..., 'argn'], + -store => \&store, + -destroy => sub {print "Final value=$frog.\n"}, + } + %vinfo = $watch->Info; + $args = $watch->Args(-fetch); + $val = $watch->Fetch; + print "val=", $watch->Say($val), ".\n"; + $watch->Store('Hello'); + $watch->Unwatch; + +=head1 DESCRIPTION + +This class module binds one or more subroutines of your devising to a +Perl variable. All variables can have B<FETCH>, B<STORE> and +B<DESTROY> callbacks. Additionally, arrays can define B<CLEAR>, +B<DELETE>, B<EXISTS>, B<EXTEND>, B<FETCHSIZE>, B<POP>, B<PUSH>, +B<SHIFT>, B<SPLICE>, B<STORESIZE> and B<UNSHIFT> callbacks, and hashes +can define B<CLEAR>, B<DELETE>, B<EXISTS>, B<FIRSTKEY> and B<NEXTKEY> +callbacks. If these term are unfamiliar to you, I I<really> suggest +you read L<perltie>. + +With Tie::Watch you can: + + . alter a variable's value + . prevent a variable's value from being changed + . invoke a Perl/Tk callback when a variable changes + . trace references to a variable + +Callback format is patterned after the Perl/Tk scheme: supply either a +code reference, or, supply an array reference and pass the callback +code reference in the first element of the array, followed by callback +arguments. (See examples in the Synopsis, above.) + +Tie::Watch provides default callbacks for any that you fail to +specify. Other than negatively impacting performance, they perform +the standard action that you'd expect, so the variable behaves +"normally". Once you override a default callback, perhaps to insert +debug code like print statements, your callback normally finishes by +calling the underlying (overridden) method. But you don't have to! + +To map a tied method name to a default callback name simply lowercase +the tied method name and uppercase its first character. So FETCH +becomes Fetch, NEXTKEY becomes Nextkey, etcetera. + +Here are two callbacks for a scalar. The B<FETCH> (read) callback does +nothing other than illustrate the fact that it returns the value to +assign the variable. The B<STORE> (write) callback uppercases the +variable and returns it. In all cases the callback I<must> return the +correct read or write value - typically, it does this by invoking the +underlying method. + + my $fetch_scalar = sub { + my($self) = @_; + $self->Fetch; + }; + + my $store_scalar = sub { + my($self, $new_val) = @_; + $self->Store(uc $new_val); + }; + +Here are B<FETCH> and B<STORE> callbacks for either an array or hash. +They do essentially the same thing as the scalar callbacks, but +provide a little more information. + + my $fetch = sub { + my($self, $key) = @_; + my $val = $self->Fetch($key); + print "In fetch callback, key=$key, val=", $self->Say($val); + my $args = $self->Args(-fetch); + print ", args=('", join("', '", @$args), "')" if $args; + print ".\n"; + $val; + }; + + my $store = sub { + my($self, $key, $new_val) = @_; + my $val = $self->Fetch($key); + $new_val = uc $new_val; + $self->Store($key, $new_val); + print "In store callback, key=$key, val=", $self->Say($val), + ", new_val=", $self->Say($new_val); + my $args = $self->Args(-store); + print ", args=('", join("', '", @$args), "')" if $args; + print ".\n"; + $new_val; + }; + +In all cases, the first parameter is a reference to the Watch object, +used to invoke the following class methods. + +=head1 METHODS + +=over 4 + +=item $watch = Tie::Watch->new(-options => values); + +The watchpoint constructor method that accepts option/value pairs to +create and configure the Watch object. The only required option is +B<-variable>. + +B<-variable> is a I<reference> to a scalar, array or hash variable. + +B<-debug> (default 0) is 1 to activate debug print statements internal +to Tie::Watch. + +B<-shadow> (default 1) is 0 to disable array and hash shadowing. To +prevent infinite recursion Tie::Watch maintains parallel variables for +arrays and hashes. When the watchpoint is created the parallel shadow +variable is initialized with the watched variable's contents, and when +the watchpoint is deleted the shadow variable is copied to the original +variable. Thus, changes made during the watch process are not lost. +Shadowing is on my default. If you disable shadowing any changes made +to an array or hash are lost when the watchpoint is deleted. + +Specify any of the following relevant callback parameters, in the +format described above: B<-fetch>, B<-store>, B<-destroy>. +Additionally for arrays: B<-clear>, B<-extend>, B<-fetchsize>, +B<-pop>, B<-push>, B<-shift>, B<-splice>, B<-storesize> and +B<-unshift>. Additionally for hashes: B<-clear>, B<-delete>, +B<-exists>, B<-firstkey> and B<-nextkey>. + +=item $args = $watch->Args(-fetch); + +Returns a reference to a list of arguments for the specified callback, +or undefined if none. + +=item $watch->Fetch(); $watch->Fetch($key); + +Returns a variable's current value. $key is required for an array or +hash. + +=item %vinfo = $watch->Info(); + +Returns a hash detailing the internals of the Watch object, with these +keys: + + %vinfo = { + -variable => SCALAR(0x200737f8) + -debug => '0' + -shadow => '1' + -value => 'HELLO SCALAR' + -destroy => ARRAY(0x200f86cc) + -fetch => ARRAY(0x200f8558) + -store => ARRAY(0x200f85a0) + -legible => above data formatted as a list of string, for printing + } + +For array and hash Watch objects, the B<-value> key is replaced with a +B<-ptr> key which is a reference to the parallel array or hash. +Additionally, for an array or hash, there are key/value pairs for +all the variable specific callbacks. + +=item $watch->Say($val); + +Used mainly for debugging, it returns $val in quotes if required, or +the string "undefined" for undefined values. + +=item $watch->Store($new_val); $watch->Store($key, $new_val); + +Store a variable's new value. $key is required for an array or hash. + +=item $watch->Unwatch(); + +Stop watching the variable. + +=back + +=head1 EFFICIENCY CONSIDERATIONS + +If you can live using the class methods provided, please do so. You +can meddle with the object hash directly and improved watch +performance, at the risk of your code breaking in the future. + +=head1 AUTHOR + +Stephen O. Lidie + +=head1 HISTORY + + lusol@Lehigh.EDU, LUCC, 96/05/30 + . Original version 0.92 release, based on the Trace module from Hans Mulder, + and ideas from Tim Bunce. + + lusol@Lehigh.EDU, LUCC, 96/12/25 + . Version 0.96, release two inner references detected by Perl 5.004. + + lusol@Lehigh.EDU, LUCC, 97/01/11 + . Version 0.97, fix Makefile.PL and MANIFEST (thanks Andreas Koenig). + Make sure test.pl doesn't fail if Tk isn't installed. + + Stephen.O.Lidie@Lehigh.EDU, Lehigh University Computing Center, 97/10/03 + . Version 0.98, implement -shadow option for arrays and hashes. + + Stephen.O.Lidie@Lehigh.EDU, Lehigh University Computing Center, 98/02/11 + . Version 0.99, finally, with Perl 5.004_57, we can completely watch arrays. + With tied array support this module is essentially complete, so its been + optimized for speed at the expense of clarity - sorry about that. The + Delete() method has been renamed Unwatch() because it conflicts with the + builtin delete(). + + Stephen.O.Lidie@Lehigh.EDU, Lehigh University Computing Center, 99/04/04 + . Version 1.0, for Perl 5.005_03, update Makefile.PL for ActiveState, and + add two examples (one for Perl/Tk). + + sol0@lehigh.edu, Lehigh University Computing Center, 2003/06/07 + . Version 1.1, for Perl 5.8, can trace a reference now, patch from Slaven + Rezic. + + sol0@lehigh.edu, Lehigh University Computing Center, 2005/05/17 + . Version 1.2, for Perl 5.8, per Rob Seegel's suggestion, support array + DELETE and EXISTS. + +=head1 COPYRIGHT + +Copyright (C) 1996 - 2005 Stephen O. Lidie. All rights reserved. + +This program is free software; you can redistribute it and/or modify it under +the same terms as Perl itself. + +=cut + +use 5.004_57;; +use Carp; +use strict; +use subs qw/normalize_callbacks/; +use vars qw/@array_callbacks @hash_callbacks @scalar_callbacks/; + +@array_callbacks = qw/-clear -delete -destroy -exists -extend -fetch + -fetchsize -pop -push -shift -splice -store + -storesize -unshift/; +@hash_callbacks = qw/-clear -delete -destroy -exists -fetch -firstkey + -nextkey -store/; +@scalar_callbacks = qw/-destroy -fetch -store/; + +sub new { + + # Watch constructor. The *real* constructor is Tie::Watch->base_watch(), + # invoked by methods in other Watch packages, depending upon the variable's + # type. Here we supply defaulted parameter values and then verify them, + # normalize all callbacks and bind the variable to the appropriate package. + + my($class, %args) = @_; + my $version = $Tie::Watch::VERSION; + my (%arg_defaults) = (-debug => 0, -shadow => 1); + my $variable = $args{-variable}; + croak "Tie::Watch::new(): -variable is required." if not defined $variable; + + my($type, $watch_obj) = (ref $variable, undef); + if ($type =~ /(SCALAR|REF)/) { + @arg_defaults{@scalar_callbacks} = ( + [\&Tie::Watch::Scalar::Destroy], [\&Tie::Watch::Scalar::Fetch], + [\&Tie::Watch::Scalar::Store]); + } elsif ($type =~ /ARRAY/) { + @arg_defaults{@array_callbacks} = ( + [\&Tie::Watch::Array::Clear], [\&Tie::Watch::Array::Delete], + [\&Tie::Watch::Array::Destroy], [\&Tie::Watch::Array::Exists], + [\&Tie::Watch::Array::Extend], [\&Tie::Watch::Array::Fetch], + [\&Tie::Watch::Array::Fetchsize], [\&Tie::Watch::Array::Pop], + [\&Tie::Watch::Array::Push], [\&Tie::Watch::Array::Shift], + [\&Tie::Watch::Array::Splice], [\&Tie::Watch::Array::Store], + [\&Tie::Watch::Array::Storesize], [\&Tie::Watch::Array::Unshift]); + } elsif ($type =~ /HASH/) { + @arg_defaults{@hash_callbacks} = ( + [\&Tie::Watch::Hash::Clear], [\&Tie::Watch::Hash::Delete], + [\&Tie::Watch::Hash::Destroy], [\&Tie::Watch::Hash::Exists], + [\&Tie::Watch::Hash::Fetch], [\&Tie::Watch::Hash::Firstkey], + [\&Tie::Watch::Hash::Nextkey], [\&Tie::Watch::Hash::Store]); + } else { + croak "Tie::Watch::new() - not a variable reference."; + } + my(@margs, %ahsh, $args, @args); + @margs = grep ! defined $args{$_}, keys %arg_defaults; + %ahsh = %args; # argument hash + @ahsh{@margs} = @arg_defaults{@margs}; # fill in missing values + normalize_callbacks \%ahsh; + + if ($type =~ /(SCALAR|REF)/) { + $watch_obj = tie $$variable, 'Tie::Watch::Scalar', %ahsh; + } elsif ($type =~ /ARRAY/) { + $watch_obj = tie @$variable, 'Tie::Watch::Array', %ahsh; + } elsif ($type =~ /HASH/) { + $watch_obj = tie %$variable, 'Tie::Watch::Hash', %ahsh; + } + $watch_obj; + +} # end new, Watch constructor + +sub Args { + + # Return a reference to a list of callback arguments, or undef if none. + # + # $_[0] = self + # $_[1] = callback type + + defined $_[0]->{$_[1]}->[1] ? [@{$_[0]->{$_[1]}}[1 .. $#{$_[0]->{$_[1]}}]] + : undef; + +} # end Args + +sub Info { + + # Info() method subclassed by other Watch modules. + # + # $_[0] = self + # @_[1 .. $#_] = optional callback types + + my(%vinfo, @results); + my(@info) = (qw/-variable -debug -shadow/); + push @info, @_[1 .. $#_] if scalar @_ >= 2; + foreach my $type (@info) { + push @results, sprintf('%-10s: ', substr $type, 1) . + $_[0]->Say($_[0]->{$type}); + $vinfo{$type} = $_[0]->{$type}; + } + $vinfo{-legible} = [@results]; + %vinfo; + +} # end Info + +sub Say { + + # For debugging, mainly. + # + # $_[0] = self + # $_[1] = value + + defined $_[1] ? (ref($_[1]) ne '' ? $_[1] : "'$_[1]'") : "undefined"; + +} # end Say + +sub Unwatch { + + # Stop watching a variable by releasing the last reference and untieing it. + # Update the original variable with its shadow, if appropriate. + # + # $_[0] = self + + my $variable = $_[0]->{-variable}; + my $type = ref $variable; + my $copy = $_[0]->{-ptr} if $type !~ /(SCALAR|REF)/; + my $shadow = $_[0]->{-shadow}; + undef $_[0]; + if ($type =~ /(SCALAR|REF)/) { + untie $$variable; + } elsif ($type =~ /ARRAY/) { + untie @$variable; + @$variable = @$copy if $shadow; + } elsif ($type =~ /HASH/) { + untie %$variable; + %$variable = %$copy if $shadow; + } else { + croak "Tie::Watch::Delete() - not a variable reference."; + } + +} # end Unwatch + +# Watch private methods. + +sub base_watch { + + # Watch base class constructor invoked by other Watch modules. + + my($class, %args) = @_; + my $watch_obj = {%args}; + $watch_obj; + +} # end base_watch + +sub callback { + + # Execute a Watch callback, either the default or user specified. + # Note that the arguments are those supplied by the tied method, + # not those (if any) specified by the user when the watch object + # was instantiated. This is for performance reasons, and why the + # Args() method exists. + # + # $_[0] = self + # $_[1] = callback type + # $_[2] through $#_ = tied arguments + + &{$_[0]->{$_[1]}->[0]} ($_[0], @_[2 .. $#_]); + +} # end callback + +sub normalize_callbacks { + + # Ensure all callbacks are normalized in [\&code, @args] format. + + my($args_ref) = @_; + my($cb, $ref); + foreach my $arg (keys %$args_ref) { + next if $arg =~ /variable|debug|shadow/; + $cb = $args_ref->{$arg}; + $ref = ref $cb; + if ($ref =~ /CODE/) { + $args_ref->{$arg} = [$cb]; + } elsif ($ref !~ /ARRAY/) { + croak "Tie::Watch: malformed callback $arg=$cb."; + } + } + +} # end normalize_callbacks + +############################################################################### + +package Tie::Watch::Scalar; + +use Carp; +@Tie::Watch::Scalar::ISA = qw/Tie::Watch/; + +sub TIESCALAR { + + my($class, %args) = @_; + my $variable = $args{-variable}; + my $watch_obj = Tie::Watch->base_watch(%args); + $watch_obj->{-value} = $$variable; + print "WatchScalar new: $variable created, \@_=", join(',', @_), "!\n" + if $watch_obj->{-debug}; + bless $watch_obj, $class; + +} # end TIESCALAR + +sub Info {$_[0]->SUPER::Info('-value', @Tie::Watch::scalar_callbacks)} + +# Default scalar callbacks. + +sub Destroy {undef %{$_[0]}} +sub Fetch {$_[0]->{-value}} +sub Store {$_[0]->{-value} = $_[1]} + +# Scalar access methods. + +sub DESTROY {$_[0]->callback('-destroy')} +sub FETCH {$_[0]->callback('-fetch')} +sub STORE {$_[0]->callback('-store', $_[1])} + +############################################################################### + +package Tie::Watch::Array; + +use Carp; +@Tie::Watch::Array::ISA = qw/Tie::Watch/; + +sub TIEARRAY { + + my($class, %args) = @_; + my($variable, $shadow) = @args{-variable, -shadow}; + my @copy = @$variable if $shadow; # make a private copy of user's array + $args{-ptr} = $shadow ? \@copy : []; + my $watch_obj = Tie::Watch->base_watch(%args); + print "WatchArray new: $variable created, \@_=", join(',', @_), "!\n" + if $watch_obj->{-debug}; + bless $watch_obj, $class; + +} # end TIEARRAY + +sub Info {$_[0]->SUPER::Info('-ptr', @Tie::Watch::array_callbacks)} + +# Default array callbacks. + +sub Clear {$_[0]->{-ptr} = ()} +sub Delete {delete $_[0]->{-ptr}->[$_[1]]} +sub Destroy {undef %{$_[0]}} +sub Exists {exists $_[0]->{-ptr}->[$_[1]]} +sub Extend {} +sub Fetch {$_[0]->{-ptr}->[$_[1]]} +sub Fetchsize {scalar @{$_[0]->{-ptr}}} +sub Pop {pop @{$_[0]->{-ptr}}} +sub Push {push @{$_[0]->{-ptr}}, @_[1 .. $#_]} +sub Shift {shift @{$_[0]->{-ptr}}} +sub Splice { + my $n = scalar @_; # splice() is wierd! + return splice @{$_[0]->{-ptr}}, $_[1] if $n == 2; + return splice @{$_[0]->{-ptr}}, $_[1], $_[2] if $n == 3; + return splice @{$_[0]->{-ptr}}, $_[1], $_[2], @_[3 .. $#_] if $n >= 4; +} +sub Store {$_[0]->{-ptr}->[$_[1]] = $_[2]} +sub Storesize {$#{$_[0]->{-ptr}} = $_[1] - 1} +sub Unshift {unshift @{$_[0]->{-ptr}}, @_[1 .. $#_]} + +# Array access methods. + +sub CLEAR {$_[0]->callback('-clear')} +sub DELETE {$_[0]->callback('-delete', $_[1])} +sub DESTROY {$_[0]->callback('-destroy')} +sub EXISTS {$_[0]->callback('-exists', $_[1])} +sub EXTEND {$_[0]->callback('-extend', $_[1])} +sub FETCH {$_[0]->callback('-fetch', $_[1])} +sub FETCHSIZE {$_[0]->callback('-fetchsize')} +sub POP {$_[0]->callback('-pop')} +sub PUSH {$_[0]->callback('-push', @_[1 .. $#_])} +sub SHIFT {$_[0]->callback('-shift')} +sub SPLICE {$_[0]->callback('-splice', @_[1 .. $#_])} +sub STORE {$_[0]->callback('-store', $_[1], $_[2])} +sub STORESIZE {$_[0]->callback('-storesize', $_[1])} +sub UNSHIFT {$_[0]->callback('-unshift', @_[1 .. $#_])} + +############################################################################### + +package Tie::Watch::Hash; + +use Carp; +@Tie::Watch::Hash::ISA = qw/Tie::Watch/; + +sub TIEHASH { + + my($class, %args) = @_; + my($variable, $shadow) = @args{-variable, -shadow}; + my %copy = %$variable if $shadow; # make a private copy of user's hash + $args{-ptr} = $shadow ? \%copy : {}; + my $watch_obj = Tie::Watch->base_watch(%args); + print "WatchHash new: $variable created, \@_=", join(',', @_), "!\n" + if $watch_obj->{-debug}; + bless $watch_obj, $class; + +} # end TIEHASH + +sub Info {$_[0]->SUPER::Info('-ptr', @Tie::Watch::hash_callbacks)} + +# Default hash callbacks. + +sub Clear {$_[0]->{-ptr} = ()} +sub Delete {delete $_[0]->{-ptr}->{$_[1]}} +sub Destroy {undef %{$_[0]}} +sub Exists {exists $_[0]->{-ptr}->{$_[1]}} +sub Fetch {$_[0]->{-ptr}->{$_[1]}} +sub Firstkey {my $c = keys %{$_[0]->{-ptr}}; each %{$_[0]->{-ptr}}} +sub Nextkey {each %{$_[0]->{-ptr}}} +sub Store {$_[0]->{-ptr}->{$_[1]} = $_[2]} + +# Hash access methods. + +sub CLEAR {$_[0]->callback('-clear')} +sub DELETE {$_[0]->callback('-delete', $_[1])} +sub DESTROY {$_[0]->callback('-destroy')} +sub EXISTS {$_[0]->callback('-exists', $_[1])} +sub FETCH {$_[0]->callback('-fetch', $_[1])} +sub FIRSTKEY {$_[0]->callback('-firstkey')} +sub NEXTKEY {$_[0]->callback('-nextkey')} +sub STORE {$_[0]->callback('-store', $_[1], $_[2])} + +1; |