Bug #32732 for YAML-Tiny: [PATCH] Support for tied hash (thus ordered data)

> > Mon Jan 28 06:49:08 2008: Request 32732 was acted upon. > Transaction: Ticket created by DAMS > Queue: YAML-Tiny > Subject: [PATCH] Support for tied hash (thus ordered data) > Broken in: (no value) > Severity: Normal > Owner: Nobody > Requestors: DAMS@cpan.org > Status: new > Ticket <URL: http://rt.cpan.org/Ticket/Display.html?id=32732 > > > > Hi. > > This patch allows to use a provided module name to create tied hashes and use them when > loading a yaml string. That way, the data order can be preserved. The pach includes more POD > doc. > > > > > Only in YAML-Tiny-1.25: Makefile > Only in YAML-Tiny-1.25: blib > diff -ru YAML-Tiny-1.25_old/lib/YAML/Tiny.pm YAML-Tiny-1.25/lib/YAML/Tiny.pm > --- YAML-Tiny-1.25_old/lib/YAML/Tiny.pm 2008-01-28 16:39:05.000000000 +0700 > +++ YAML-Tiny-1.25/lib/YAML/Tiny.pm 2008-01-28 18:45:54.000000000 +0700 > @@ -14,6 +14,18 @@ > }; > } > > +our $TieHashModule = undef; > + > +# used to create a hashref, possibly using hash ties (for instance, to get ordered hash) > +sub _create_hashref { > + $TieHashModule or return {}; > + eval ("require $TieHashModule"); > + $@ and die "YAML::Tiny failed requiring '$TieHashModule', specified to be used as a hash tie"; > + my %hash; > + tie %hash, $TieHashModule; > + return \%hash; > +} > + > my $ESCAPE_CHAR = '[\\x00-\\x08\\x0b-\\x0d\\x0e-\\x1f\"\n]'; > > # Escapes for unprintable characters > @@ -100,7 +112,7 @@ > > } elsif ( $lines[0] =~ /^(\s*)\w/ ) { > # A hash at the root > - my $document = { }; > + my $document = _create_hashref(); > push @$self, $document; > $self->_read_hash( $document, [ length($1) ], \@lines ); > > @@ -196,7 +208,7 @@ > # Inline nested hash > my $indent2 = length("$1"); > $lines->[0] =~ s/-/ /; > - push @$array, { }; > + push @$array, _create_hashref(); > $self->_read_hash( $array->[-1], [ @$indent, $indent2 ], $lines ); > > } elsif ( $lines->[0] =~ /^\s*\-(\s*)(.+?)\s*$/ ) { > @@ -222,7 +234,7 @@ > } > > } elsif ( $lines->[0] =~ /^(\s*)\w/ ) { > - push @$array, { }; > + push @$array, _create_hashref(); > $self->_read_hash( $array->[-1], [ @$indent, length("$1") ], $lines ); > > } else { > @@ -279,7 +291,7 @@ > # Null hash entry > $hash->{$key} = undef; > } else { > - $hash->{$key} = {}; > + $hash->{$key} = _create_hashref(); > $self->_read_hash( $hash->{$key}, [ @$indent, length($1) ], $lines ); > } > } > @@ -411,7 +423,7 @@ > die "YAML::Tiny does not support circular references"; > } > my @lines = (); > - foreach my $name ( sort keys %$hash ) { > + foreach my $name ( $TieHashModule ? sort keys %$hash : keys %$hash ) { > my $el = $hash->{$name}; > my $line = (' ' x $indent) . "$name:"; > my $type = ref $el; > @@ -608,8 +620,8 @@ > sense of. > > L<YAML::Tiny> does not generate comments, it won't necesarily preserve the > -order of your hashes, and it will normalise if reading in and writing out > -again. > +order of your hashes (except if you use C<YAML::Tiny::TieHashModule>, see > +below), and it will normalise if reading in and writing out again. > > It only supports a very basic subset of the full YAML specification. > > @@ -992,6 +1004,19 @@ > > Reads the YAML stream from a file instead of a string. > > +=head1 VARIABLES > + > +=head2 C<YAML::Tiny::TieHashModule> > + > +This module variable can be use to specify a module to be used to ie hashes > +against, thus providing more features to the hashes. For instance, using > +Tie::IxHash, Tie::Hash::Indexed or Ties::Hash::Sorted, the order of YAML data > +can be preserved, or sorted. The following example rounds-trip better : > + > + local $YAML::Tiny::TieHashModule = 'Tie::IxHash'; > + my ($structure) = YAML::Tiny::LoadFile($filename); > + my $output = YAML::Tiny::DumpFile($filename); > + > =head1 SUPPORT > > Bugs should be reported via the CPAN bug tracker at > Only in YAML-Tiny-1.25: pm_to_blib > >

> > Queue: YAML-Tiny > Ticket <URL: http://rt.cpan.org/Ticket/Display.html?id=32732 > > > I agree and fully understand your position. I had the same opinion and wanted to create > YAML::Tiny::Tie (or similar name) at the first place, but then I thought you could find it impolite > to do it so without having your word to say. > > However, as we agree, do you mind if I release YAML::Tiny::Tie soon ? > > Thanks, > dams >

> Only in YAML-Tiny-1.25: Makefile > Only in YAML-Tiny-1.25: blib > diff -ru YAML-Tiny-1.25_old/lib/YAML/Tiny.pm YAML-Tiny-1.25/lib/YAML/Tiny.pm > --- YAML-Tiny-1.25_old/lib/YAML/Tiny.pm 2008-01-28 16:39:05.000000000 +0700 > +++ YAML-Tiny-1.25/lib/YAML/Tiny.pm 2008-01-29 11:36:04.000000000 +0700 > @@ -14,6 +14,13 @@ > }; > } > > +our $DontForceAzOrdering = undef; > + > +# used to create new hashrefs and arrayrefs. Can be overloaded to use tied hash > +# and arrays, (for instance, to get ordered hash) > +sub _create_hashref { {} } > +sub _create_arrayref { [] } > + > my $ESCAPE_CHAR = '[\\x00-\\x08\\x0b-\\x0d\\x0e-\\x1f\"\n]'; > > # Escapes for unprintable characters > @@ -94,13 +101,13 @@ > > } elsif ( $lines[0] =~ /^\s*\-/ ) { > # An array at the root > - my $document = [ ]; > + my $document = $self->_create_arrayref(); > push @$self, $document; > $self->_read_array( $document, [ 0 ], \@lines ); > > } elsif ( $lines[0] =~ /^(\s*)\w/ ) { > # A hash at the root > - my $document = { }; > + my $document = $self->_create_hashref(); > push @$self, $document; > $self->_read_hash( $document, [ length($1) ], \@lines ); > > @@ -143,11 +150,11 @@ > # Null hash and array > if ( $string eq '{}' ) { > # Null hash > - return {}; > + return $self->_create_hashref(); > } > if ( $string eq '[]' ) { > # Null array > - return []; > + return $self->_create_arrayref(); > } > > # Regular unquoted string > @@ -196,7 +203,7 @@ > # Inline nested hash > my $indent2 = length("$1"); > $lines->[0] =~ s/-/ /; > - push @$array, { }; > + push @$array, $self->_create_hashref(); > $self->_read_hash( $array->[-1], [ @$indent, $indent2 ], $lines ); > > } elsif ( $lines->[0] =~ /^\s*\-(\s*)(.+?)\s*$/ ) { > @@ -217,12 +224,12 @@ > push @$array, undef; > } else { > # Naked indenter > - push @$array, [ ]; > + push @$array, $self->_create_arrayref(); > $self->_read_array( $array->[-1], [ @$indent, $indent2 ], $lines ); > } > > } elsif ( $lines->[0] =~ /^(\s*)\w/ ) { > - push @$array, { }; > + push @$array, $self->_create_hashref(); > $self->_read_hash( $array->[-1], [ @$indent, length("$1") ], $lines ); > > } else { > @@ -271,7 +278,7 @@ > return 1; > } > if ( $lines->[0] =~ /^(\s*)-/ ) { > - $hash->{$key} = []; > + $hash->{$key} = $self->_create_arrayref(); > $self->_read_array( $hash->{$key}, [ @$indent, length($1) ], $lines ); > } elsif ( $lines->[0] =~ /^(\s*)./ ) { > my $indent2 = length("$1"); > @@ -279,7 +286,7 @@ > # Null hash entry > $hash->{$key} = undef; > } else { > - $hash->{$key} = {}; > + $hash->{$key} = $self->_create_hashref(); > $self->_read_hash( $hash->{$key}, [ @$indent, length($1) ], $lines ); > } > } > @@ -411,7 +418,7 @@ > die "YAML::Tiny does not support circular references"; > } > my @lines = (); > - foreach my $name ( sort keys %$hash ) { > + foreach my $name ( $DontForceAzOrdering ? keys %$hash : sort keys %$hash ) { > my $el = $hash->{$name}; > my $line = (' ' x $indent) . "$name:"; > my $type = ref $el; > @@ -607,9 +614,12 @@ > your average manager or secretary should be able to look at and make > sense of. > > -L<YAML::Tiny> does not generate comments, it won't necesarily preserve the > -order of your hashes, and it will normalise if reading in and writing out > -again. > +By default, L<YAML::Tiny> does not generate comments, it won't necesarily > +preserve the order of your hashes and it will normalise if reading in and > +writing out again. However you can alter this behaviour : see > +C<YAML::Tiny::DontForceAzOrdering> to stop forcing alphabetical reordering of > +hashes on output, and see the I<Overload> section to achieve hashes order > +preservation. > > It only supports a very basic subset of the full YAML specification. > > @@ -992,6 +1002,56 @@ > > Reads the YAML stream from a file instead of a string. > > +=head1 OVERLOADING > + > +When loading YAML content, C<YAML::Tiny> returns a new complex data structure. > +To give flexibility on this data structure creation, the following methods are > +called when creating hashes and arrays. > + > +you might want to see also C<YAML::Tiny::DontForceAzOrdering> and L<YAML::Tiny::Ordered>. > + > +=head2 C<_create_hashref> > + > +This method is used when a new hashref is needed to create the new data > +structure or one of its sub component. By default, this method simply returns > +an empty hashref {}. However, you could overload it, use tied hashes and thus > +keep the original hash orders : > + > + package YAML::Tiny::MyOrdered # your own package > + use parent qw(YAML::Tiny); # inheritance > + use Tie::IxHash; # use ordered hash > + > + sub _create_hashref { > + my %hash; > + tie %hash, 'Tie::IxHash'; > + return \%hash; > + } > + > +=head2 C<_create_arrayref> > + > +This method is used when a new arrayref is needed to create the new data > +structure or one of its sub component. By default, this method simply returns > +an empty arrayref []. However, you could overload it, and for instance use tied > +arrays. > + > +=head1 VARIABLES > + > +=head2 C<YAML::Tiny::DontForceAzOrdering> > + > +By default, when serializing a structure to a string/file, C<YAML::Tiny> > +reorder the hashes by alphabetical order. If you use overloading to use your > +own hash tie, you probably want to disable this feature. Do so by setting > +C<YAML::Tiny::DontForceAzOrdering> to a true value when using C<write>, > +C<write_string>, C<Dump> or C<DumpFile>. That way, the hash keys order will be > +the one return by the C<keys> function. > + > + my ($structure) = YAML::Tiny::LoadFile($filename); > + #... > + { > + local $YAML::Tiny::DontForceAzOrdering = 1; > + my $output = YAML::Tiny::DumpFile($filename); > + } > + > =head1 SUPPORT > > Bugs should be reported via the CPAN bug tracker at > Only in YAML-Tiny-1.25: pm_to_blib > >