[scpubgit/stemmatology.git] / base / lib / Text / Tradition.pm

package Text::Tradition;

use JSON qw / from_json /;
use Module::Load;
use Moose;
use Moose::Util qw/ does_role apply_all_roles /;
use Text::Tradition::Collation;
use Text::Tradition::Error;
use Text::Tradition::Witness;
use TryCatch;

use vars qw( $VERSION );
$VERSION = "1.1";

# Enable plugin(s) if available
eval { with 'Text::Tradition::HasStemma'; };
# Don't warn normally
# if( $@ ) {
# 	warn "Text::Tradition::Analysis not found. Disabling stemma analysis functionality";
# };
eval { with 'Text::Tradition::Language'; };
eval { with 'Text::Tradition::Ownership'; };

has 'collation' => (
    is => 'ro',
    isa => 'Text::Tradition::Collation',
    writer => '_save_collation',
    );

has 'witness_hash' => (
    traits => ['Hash'],
    isa => 'HashRef[Text::Tradition::Witness]',
    handles => {
        witness     => 'get',
        add_witness => 'set',
        del_witness => 'delete',
        has_witness => 'exists',
        witnesses   => 'values',
    },
    default => sub { {} },
    );

has 'name' => (
    is => 'rw',
    isa => 'Str',
    default => 'Tradition',
    );
    
has '_initialized' => (
	is => 'ro',
	isa => 'Bool',
	default => undef,
	writer => '_init_done',
	); 

# Create the witness before trying to add it
around 'add_witness' => sub {
    my $orig = shift;
    my $self = shift;
    # TODO allow add of a Witness object?
    my %args = @_ == 1 ? %{$_[0]} : @_;
    $args{'tradition'} = $self;
    my $new_wit = Text::Tradition::Witness->new( %args );
    $self->$orig( $new_wit->sigil => $new_wit );
    return $new_wit;
};

# Allow deletion of witness by object as well as by sigil
around 'del_witness' => sub {
    my $orig = shift;
    my $self = shift;
    my @key_args;
    foreach my $arg ( @_ ) {
        push( @key_args, 
              ref( $arg ) eq 'Text::Tradition::Witness' ? $arg->sigil : $arg );
    }
    return $self->$orig( @key_args );
};

# Don't allow an empty hash value
around 'witness' => sub {
    my( $orig, $self, $arg ) = @_;
    return unless $self->has_witness( $arg );
    return $self->$orig( $arg );
};

=head1 NAME

Text::Tradition - a software model for a set of collated texts

=head1 SYNOPSIS

  use Text::Tradition;
  my $t = Text::Tradition->new( 
    'name' => 'this is a text',
    'input' => 'TEI',
    'file' => '/path/to/tei_parallel_seg_file.xml' );

  my @text_wits = $t->witnesses();
  my $manuscript_a = $t->witness( 'A' );

  $t = Text::Tradition->new();
  $t->add_witness( 'sourcetype' => 'xmldesc', 
    'file' => '/path/to/teitranscription.xml' );
  $t->add_witness( 'sourcetype => 'plaintext', 'sigil' => 'Q',
    'string' => 'The quick brown fox jumped over the lazy dogs' );
  ## TODO
  $t->collate_texts;
  
  my $text_path_svg = $t->collation->as_svg();
  ## See Text::Tradition::Collation for more on text collation itself
    
=head1 DESCRIPTION

Text::Tradition is a library for representation and analysis of collated
texts, particularly medieval ones.  A 'tradition' refers to the aggregation
of surviving versions of a text, generally preserved in multiple
manuscripts (or 'witnesses').  A Tradition object thus has one more more
Witnesses, as well as a Collation that represents the unity of all versions
of the text.

=head1 METHODS

=head2 new

Creates and returns a new text tradition object.  The following options are
accepted.

General options:

=over 4

=item B<name> - The name of the text.

=back

Initialization based on a collation file:

=over 4

=item B<input> - The input format of the collation file.  Can be one of the
following:

=over 4

=item * Self - a GraphML format produced by this module

=item * CollateX - a GraphML format produced by CollateX

=item * CTE - a TEI XML format produced by Classical Text Editor

=item * JSON - an alignment table in JSON format, as produced by CollateX and 
other tools

=item * TEI - a TEI parallel segmentation format file

=item * Tabular - a spreadsheet collation.  See the documentation for 
L<Text::Tradition::Parser::Tabular> for an explanation of additional options.

=back

=item B<file> - The name of the file that contains the data.  One of 'file'
or 'string' should be specified.

=item B<string> - A text string that contains the data.  One of 'file' or
'string' should be specified.

=back

Initialization based on a list of witnesses [NOT YET IMPLEMENTED]:

=over 4

=item B<witnesses> - A reference to an array of Text::Tradition::Witness
objects that carry the text to be collated.

=item B<collator> - A reference to a collation program that will accept
Witness objects.

=back

=head2 B<witnesses>

Return the Text::Tradition::Witness objects associated with this tradition,
as an array.

=head2 B<witness>( $sigil )

Returns the Text::Tradition::Witness object whose sigil is $sigil, or undef
if there is no such object within the tradition.

=head2 B<add_witness>( %opts )

Instantiate a new witness with the given options (see documentation for
Text::Tradition::Witness) and add it to the tradition.

=head2 B<del_witness>( $sigil )

Delete the witness with the given sigil from the tradition.  Returns the
witness object for the deleted witness.

=begin testing

use_ok( 'Text::Tradition', "can use module" );

my $t = Text::Tradition->new( 'name' => 'empty' );
is( ref( $t ), 'Text::Tradition', "initialized an empty Tradition object" );
is( $t->name, 'empty', "object has the right name" );
is( scalar $t->witnesses, 0, "object has no witnesses" );

my $simple = 't/data/simple.txt';
my $s = Text::Tradition->new( 
    'name'  => 'inline', 
    'input' => 'Tabular',
    'file'  => $simple,
    );
is( ref( $s ), 'Text::Tradition', "initialized a Tradition object" );
is( $s->name, 'inline', "object has the right name" );
is( scalar $s->witnesses, 3, "object has three witnesses" );

my $wit_a = $s->witness('A');
is( ref( $wit_a ), 'Text::Tradition::Witness', "Found a witness A" );
if( $wit_a ) {
    is( $wit_a->sigil, 'A', "Witness A has the right sigil" );
}
is( $s->witness('X'), undef, "There is no witness X" );
ok( !exists $s->{'witnesses'}->{'X'}, "Witness key X not created" );

my $wit_d = $s->add_witness( 'sigil' => 'D', 'sourcetype' => 'collation' );
is( ref( $wit_d ), 'Text::Tradition::Witness', "new witness created" );
is( $wit_d->sigil, 'D', "witness has correct sigil" );
is( scalar $s->witnesses, 4, "object now has four witnesses" );

my $del = $s->del_witness( 'D' );
is( $del, $wit_d, "Deleted correct witness" );
is( scalar $s->witnesses, 3, "object has three witnesses again" );

# TODO test initialization by witness list when we have it

=end testing

=cut
    

sub BUILD {
    my( $self, $init_args ) = @_;
    
    # First, make a collation object. This will use only those arguments in
    # init_args that apply to the collation.
	my $collation = Text::Tradition::Collation->new( %$init_args,
													'tradition' => $self );
	$self->_save_collation( $collation );

    if( exists $init_args->{'input'} ) {
        # Call the appropriate parser on the given data
        my @format_standalone = qw/ Self CollateText CollateX CTE JSON TEI Tabular /;
        my @format_basetext = qw/ KUL /;
        my $use_base;
        my $format = $init_args->{'input'};
        if( $format && !( grep { $_ eq $format } @format_standalone )
            && !( grep { $_ eq $format } @format_basetext ) ) {
            warn "Unrecognized input format $format; not parsing";
            return;
        }
        if( $format && grep { $_ eq $format } @format_basetext ) {
            $use_base = 1;
            if( !exists $init_args->{'base'} ) {
                warn "Cannot make a collation from $format without a base text";
                return;
            }
        }

        # Now do the parsing. 
        if( $format ) {
            if( $use_base ) { 
                $format = 'BaseText';   # Use the BaseText module for parsing,
                                        # but retain the original input arg.
            }
            my $mod = "Text::Tradition::Parser::$format";
            load( $mod );
            $mod->can('parse')->( $self, $init_args );
        }
    }
    $self->_init_done( 1 );
    return $self;
}

=head2 add_json_witnesses( $jsonstring, $options )

Adds a set of witnesses from a JSON array specification. This is a wrapper
to parse the JSON and call add_witness (with the specified $options) for
each element therein.

=cut

sub add_json_witnesses {
	my( $self, $jsonstr, $extraopts ) = @_;
	my $witarray = from_json( $jsonstr );
	foreach my $witspec ( @{$witarray->{witnesses}} ) {
		my $opts = $extraopts || {};
		$opts->{'sourcetype'} = 'json';
		$opts->{'object'} = $witspec;
		$self->add_witness( $opts );
	}
}

sub throw {
	my $self = shift;
	Text::Tradition::Error->throw( 
		'ident' => 'Tradition error',
		'message' => $_[0],
		);
}

no Moose;
__PACKAGE__->meta->make_immutable;


=head1 BUGS / TODO

=over

=item * Allow tradition to be initialized via passing to a collator.

=back

=head1 LICENSE

This package is free software and is provided "as is" without express
or implied warranty.  You can redistribute it and/or modify it under
the same terms as Perl itself.

=head1 AUTHOR

Tara L Andrews E<lt>aurum@cpan.orgE<gt>
Commit	Line	Data
dd3b58b0	1	package Text::Tradition;
dd3b58b0	2
65ed66b9	3	use JSON qw / from_json /;
4a8828f0	4	use Module::Load;
dd3b58b0	5	use Moose;
951ddfe8	6	use Moose::Util qw/ does_role apply_all_roles /;
8e1394aa	7	use Text::Tradition::Collation;
951ddfe8	8	use Text::Tradition::Error;
8e1394aa	9	use Text::Tradition::Witness;
951ddfe8	10	use TryCatch;
dd3b58b0	11
331c2dbf	12	use vars qw( $VERSION );
37bf09f4	13	$VERSION = "1.1";
	14
	15	# Enable plugin(s) if available
	16	eval { with 'Text::Tradition::HasStemma'; };
ed5b9b70	17	# Don't warn normally
	18	# if( $@ ) {
	19	# warn "Text::Tradition::Analysis not found. Disabling stemma analysis functionality";
	20	# };
e92d4229	21	eval { with 'Text::Tradition::Language'; };
8943ff68	22	eval { with 'Text::Tradition::Ownership'; };
331c2dbf	23
dd3b58b0	24	has 'collation' => (
8e1394aa	25	is => 'ro',
	26	isa => 'Text::Tradition::Collation',
	27	writer => '_save_collation',
	28	);
dd3b58b0	29
3b853983	30	has 'witness_hash' => (
	31	traits => ['Hash'],
	32	isa => 'HashRef[Text::Tradition::Witness]',
8e1394aa	33	handles => {
3b853983	34	witness => 'get',
	35	add_witness => 'set',
	36	del_witness => 'delete',
	37	has_witness => 'exists',
	38	witnesses => 'values',
8e1394aa	39	},
3b853983	40	default => sub { {} },
8e1394aa	41	);
c5104dc0	42
df6d9812	43	has 'name' => (
	44	is => 'rw',
	45	isa => 'Str',
	46	default => 'Tradition',
	47	);
56cf65bd	48
3579c22b	49	has '_initialized' => (
10943ab0	50	is => 'ro',
	51	isa => 'Bool',
	52	default => undef,
	53	writer => '_init_done',
	54	);
	55
3b853983	56	# Create the witness before trying to add it
910a0a6d	57	around 'add_witness' => sub {
	58	my $orig = shift;
	59	my $self = shift;
331c2dbf	60	# TODO allow add of a Witness object?
fae52efd	61	my %args = @_ == 1 ? %{$_[0]} : @_;
fae52efd	62	$args{'tradition'} = $self;
fae52efd	63	my $new_wit = Text::Tradition::Witness->new( %args );
3b853983	64	$self->$orig( $new_wit->sigil => $new_wit );
910a0a6d	65	return $new_wit;
910a0a6d	66	};
331c2dbf	67
3b853983	68	# Allow deletion of witness by object as well as by sigil
	69	around 'del_witness' => sub {
	70	my $orig = shift;
	71	my $self = shift;
	72	my @key_args;
	73	foreach my $arg ( @_ ) {
	74	push( @key_args,
	75	ref( $arg ) eq 'Text::Tradition::Witness' ? $arg->sigil : $arg );
	76	}
	77	return $self->$orig( @key_args );
	78	};
	79
	80	# Don't allow an empty hash value
	81	around 'witness' => sub {
	82	my( $orig, $self, $arg ) = @_;
	83	return unless $self->has_witness( $arg );
	84	return $self->$orig( $arg );
	85	};
	86
331c2dbf	87	=head1 NAME
	88
	89	Text::Tradition - a software model for a set of collated texts
	90
	91	=head1 SYNOPSIS
	92
	93	use Text::Tradition;
	94	my $t = Text::Tradition->new(
	95	'name' => 'this is a text',
	96	'input' => 'TEI',
	97	'file' => '/path/to/tei_parallel_seg_file.xml' );
	98
	99	my @text_wits = $t->witnesses();
	100	my $manuscript_a = $t->witness( 'A' );
82fa4d57	101
	102	$t = Text::Tradition->new();
	103	$t->add_witness( 'sourcetype' => 'xmldesc',
	104	'file' => '/path/to/teitranscription.xml' );
	105	$t->add_witness( 'sourcetype => 'plaintext', 'sigil' => 'Q',
	106	'string' => 'The quick brown fox jumped over the lazy dogs' );
	107	## TODO
	108	$t->collate_texts;
331c2dbf	109
	110	my $text_path_svg = $t->collation->as_svg();
	111	## See Text::Tradition::Collation for more on text collation itself
	112
	113	=head1 DESCRIPTION
	114
	115	Text::Tradition is a library for representation and analysis of collated
	116	texts, particularly medieval ones. A 'tradition' refers to the aggregation
	117	of surviving versions of a text, generally preserved in multiple
	118	manuscripts (or 'witnesses'). A Tradition object thus has one more more
	119	Witnesses, as well as a Collation that represents the unity of all versions
	120	of the text.
	121
	122	=head1 METHODS
	123
	124	=head2 new
	125
	126	Creates and returns a new text tradition object. The following options are
	127	accepted.
	128
	129	General options:
	130
	131	=over 4
	132
	133	=item B<name> - The name of the text.
	134
	135	=back
	136
	137	Initialization based on a collation file:
	138
	139	=over 4
	140
	141	=item B<input> - The input format of the collation file. Can be one of the
	142	following:
	143
	144	=over 4
	145
	146	=item * Self - a GraphML format produced by this module
	147
	148	=item * CollateX - a GraphML format produced by CollateX
	149
	150	=item * CTE - a TEI XML format produced by Classical Text Editor
	151
a445ce40	152	=item * JSON - an alignment table in JSON format, as produced by CollateX and
a445ce40	153	other tools
331c2dbf	154
	155	=item * TEI - a TEI parallel segmentation format file
	156
a445ce40	157	=item * Tabular - a spreadsheet collation. See the documentation for
a445ce40	158	L<Text::Tradition::Parser::Tabular> for an explanation of additional options.
331c2dbf	159
	160	=back
	161
	162	=item B<file> - The name of the file that contains the data. One of 'file'
	163	or 'string' should be specified.
	164
	165	=item B<string> - A text string that contains the data. One of 'file' or
	166	'string' should be specified.
	167
331c2dbf	168	=back
	169
	170	Initialization based on a list of witnesses [NOT YET IMPLEMENTED]:
	171
	172	=over 4
	173
	174	=item B<witnesses> - A reference to an array of Text::Tradition::Witness
	175	objects that carry the text to be collated.
	176
	177	=item B<collator> - A reference to a collation program that will accept
	178	Witness objects.
	179
	180	=back
	181
	182	=head2 B<witnesses>
	183
	184	Return the Text::Tradition::Witness objects associated with this tradition,
	185	as an array.
	186
044d1e45	187	=head2 B<witness>( $sigil )
	188
	189	Returns the Text::Tradition::Witness object whose sigil is $sigil, or undef
	190	if there is no such object within the tradition.
	191
331c2dbf	192	=head2 B<add_witness>( %opts )
	193
	194	Instantiate a new witness with the given options (see documentation for
	195	Text::Tradition::Witness) and add it to the tradition.
	196
044d1e45	197	=head2 B<del_witness>( $sigil )
	198
	199	Delete the witness with the given sigil from the tradition. Returns the
	200	witness object for the deleted witness.
	201
331c2dbf	202	=begin testing
	203
	204	use_ok( 'Text::Tradition', "can use module" );
	205
	206	my $t = Text::Tradition->new( 'name' => 'empty' );
	207	is( ref( $t ), 'Text::Tradition', "initialized an empty Tradition object" );
	208	is( $t->name, 'empty', "object has the right name" );
	209	is( scalar $t->witnesses, 0, "object has no witnesses" );
	210
	211	my $simple = 't/data/simple.txt';
	212	my $s = Text::Tradition->new(
	213	'name' => 'inline',
	214	'input' => 'Tabular',
	215	'file' => $simple,
	216	);
	217	is( ref( $s ), 'Text::Tradition', "initialized a Tradition object" );
	218	is( $s->name, 'inline', "object has the right name" );
	219	is( scalar $s->witnesses, 3, "object has three witnesses" );
	220
044d1e45	221	my $wit_a = $s->witness('A');
	222	is( ref( $wit_a ), 'Text::Tradition::Witness', "Found a witness A" );
	223	if( $wit_a ) {
	224	is( $wit_a->sigil, 'A', "Witness A has the right sigil" );
	225	}
	226	is( $s->witness('X'), undef, "There is no witness X" );
	227	ok( !exists $s->{'witnesses'}->{'X'}, "Witness key X not created" );
	228
82fa4d57	229	my $wit_d = $s->add_witness( 'sigil' => 'D', 'sourcetype' => 'collation' );
044d1e45	230	is( ref( $wit_d ), 'Text::Tradition::Witness', "new witness created" );
044d1e45	231	is( $wit_d->sigil, 'D', "witness has correct sigil" );
331c2dbf	232	is( scalar $s->witnesses, 4, "object now has four witnesses" );
331c2dbf	233
3b853983	234	my $del = $s->del_witness( 'D' );
044d1e45	235	is( $del, $wit_d, "Deleted correct witness" );
3b853983	236	is( scalar $s->witnesses, 3, "object has three witnesses again" );
3b853983	237
331c2dbf	238	# TODO test initialization by witness list when we have it
	239
	240	=end testing
	241
	242	=cut
910a0a6d	243
df6d9812	244
8e1394aa	245	sub BUILD {
8e1394aa	246	my( $self, $init_args ) = @_;
fae52efd	247
	248	# First, make a collation object. This will use only those arguments in
	249	# init_args that apply to the collation.
	250	my $collation = Text::Tradition::Collation->new( %$init_args,
	251	'tradition' => $self );
	252	$self->_save_collation( $collation );
c5104dc0	253
fae52efd	254	if( exists $init_args->{'input'} ) {
910a0a6d	255	# Call the appropriate parser on the given data
a731e73a	256	my @format_standalone = qw/ Self CollateText CollateX CTE JSON TEI Tabular /;
dfc37e38	257	my @format_basetext = qw/ KUL /;
	258	my $use_base;
	259	my $format = $init_args->{'input'};
dfc37e38	260	if( $format && !( grep { $_ eq $format } @format_standalone )
	261	&& !( grep { $_ eq $format } @format_basetext ) ) {
	262	warn "Unrecognized input format $format; not parsing";
910a0a6d	263	return;
910a0a6d	264	}
dfc37e38	265	if( $format && grep { $_ eq $format } @format_basetext ) {
	266	$use_base = 1;
	267	if( !exists $init_args->{'base'} ) {
	268	warn "Cannot make a collation from $format without a base text";
	269	return;
	270	}
	271	}
4a8828f0	272
910a0a6d	273	# Now do the parsing.
910a0a6d	274	if( $format ) {
dfc37e38	275	if( $use_base ) {
	276	$format = 'BaseText'; # Use the BaseText module for parsing,
	277	# but retain the original input arg.
910a0a6d	278	}
	279	my $mod = "Text::Tradition::Parser::$format";
	280	load( $mod );
dfc37e38	281	$mod->can('parse')->( $self, $init_args );
910a0a6d	282	}
c5104dc0	283	}
10943ab0	284	$self->_init_done( 1 );
fae52efd	285	return $self;
	286	}
	287
	288	=head2 add_json_witnesses( $jsonstring, $options )
	289
	290	Adds a set of witnesses from a JSON array specification. This is a wrapper
	291	to parse the JSON and call add_witness (with the specified $options) for
	292	each element therein.
	293
	294	=cut
	295
	296	sub add_json_witnesses {
	297	my( $self, $jsonstr, $extraopts ) = @_;
65ed66b9	298	my $witarray = from_json( $jsonstr );
65ed66b9	299	foreach my $witspec ( @{$witarray->{witnesses}} ) {
fae52efd	300	my $opts = $extraopts \|\| {};
	301	$opts->{'sourcetype'} = 'json';
	302	$opts->{'object'} = $witspec;
	303	$self->add_witness( $opts );
	304	}
8e1394aa	305	}
c5104dc0	306
951ddfe8	307	sub throw {
142698b8	308	my $self = shift;
951ddfe8	309	Text::Tradition::Error->throw(
	310	'ident' => 'Tradition error',
	311	'message' => $_[0],
	312	);
	313	}
	314
dd3b58b0	315	no Moose;
dd3b58b0	316	__PACKAGE__->meta->make_immutable;
331c2dbf	317
	318
	319	=head1 BUGS / TODO
	320
	321	=over
	322
	323	=item * Allow tradition to be initialized via passing to a collator.
	324
	325	=back
	326
	327	=head1 LICENSE
	328
	329	This package is free software and is provided "as is" without express
	330	or implied warranty. You can redistribute it and/or modify it under
	331	the same terms as Perl itself.
	332
	333	=head1 AUTHOR
	334
	335	Tara L Andrews E<lt>aurum@cpan.orgE<gt>