[scpubgit/stemmatology.git] / base / t / text_tradition_collation.t

#!/usr/bin/perl -w

use strict;
use Test::More 'no_plan';
$| = 1;


# =begin testing
{
use Text::Tradition;

my $cxfile = 't/data/Collatex-16.xml';
my $t = Text::Tradition->new( 
    'name'  => 'inline', 
    'input' => 'CollateX',
    'file'  => $cxfile,
    );
my $c = $t->collation;

my $rno = scalar $c->readings;
# Split n21 ('unto') for testing purposes
my $new_r = $c->add_reading( { 'id' => 'n21p0', 'text' => 'un', 'join_next' => 1 } );
my $old_r = $c->reading( 'n21' );
$old_r->alter_text( 'to' );
$c->del_path( 'n20', 'n21', 'A' );
$c->add_path( 'n20', 'n21p0', 'A' );
$c->add_path( 'n21p0', 'n21', 'A' );
$c->add_relationship( 'n21', 'n22', { type => 'collated', scope => 'local' } );
$c->flatten_ranks();
ok( $c->reading( 'n21p0' ), "New reading exists" );
is( scalar $c->readings, $rno, "Reading add offset by flatten_ranks" );

# Combine n3 and n4 ( with his )
$c->merge_readings( 'n3', 'n4', 1 );
ok( !$c->reading('n4'), "Reading n4 is gone" );
is( $c->reading('n3')->text, 'with his', "Reading n3 has both words" );

# Collapse n9 and n10 ( rood / root )
$c->merge_readings( 'n9', 'n10' );
ok( !$c->reading('n10'), "Reading n10 is gone" );
is( $c->reading('n9')->text, 'rood', "Reading n9 has an unchanged word" );

# Combine n21 and n21p0
my $remaining = $c->reading('n21');
$remaining ||= $c->reading('n22');  # one of these should still exist
$c->merge_readings( 'n21p0', $remaining, 1 );
ok( !$c->reading('n21'), "Reading $remaining is gone" );
is( $c->reading('n21p0')->text, 'unto', "Reading n21p0 merged correctly" );
}


# =begin testing
{
use Text::Tradition;

my $st = Text::Tradition->new( 'input' => 'Self', 'file' => 't/data/collatecorr.xml' );
is( ref( $st ), 'Text::Tradition', "Got a tradition from test file" );
ok( $st->has_witness('Ba96'), "Tradition has the affected witness" );

my $sc = $st->collation;
my $numr = 17;
ok( $sc->reading('n131'), "Tradition has the affected reading" );
is( scalar( $sc->readings ), $numr, "There are $numr readings in the graph" );
is( $sc->end->rank, 14, "There are fourteen ranks in the graph" );

# Detach the erroneously collated reading
my( $newr, @del_rdgs ) = $sc->duplicate_reading( 'n131', 'Ba96' );
ok( $newr, "New reading was created" );
ok( $sc->reading('n131_0'), "Detached the bad collation with a new reading" );
is( scalar( $sc->readings ), $numr + 1, "A reading was added to the graph" );
is( $sc->end->rank, 10, "There are now only ten ranks in the graph" );
my $csucc = $sc->common_successor( 'n131', 'n131_0' );
is( $csucc->id, 'n136', "Found correct common successor to duped reading" ); 

# Check that the bad transposition is gone
is( scalar @del_rdgs, 1, "Deleted reading was returned by API call" );
is( $sc->get_relationship( 'n130', 'n135' ), undef, "Bad transposition relationship is gone" );

# The collation should not be fixed
my @pairs = $sc->identical_readings();
is( scalar @pairs, 0, "Not re-collated yet" );
# Fix the collation
ok( $sc->merge_readings( 'n124', 'n131_0' ), "Collated the readings correctly" );
@pairs = $sc->identical_readings( start => 'n124', end => $csucc->id );
is( scalar @pairs, 3, "Found three more identical readings" );
is( $sc->end->rank, 11, "The ranks shifted appropriately" );
$sc->flatten_ranks();
is( scalar( $sc->readings ), $numr - 3, "Now we are collated correctly" );
}


# =begin testing
{
use Text::Tradition;
use TryCatch;

my $READINGS = 311;
my $PATHS = 361;

my $datafile = 't/data/florilegium_tei_ps.xml';
my $tradition = Text::Tradition->new( 'input' => 'TEI',
                                      'name' => 'test0',
                                      'file' => $datafile,
                                      'linear' => 1 );

ok( $tradition, "Got a tradition object" );
is( scalar $tradition->witnesses, 13, "Found all witnesses" );
ok( $tradition->collation, "Tradition has a collation" );

my $c = $tradition->collation;
is( scalar $c->readings, $READINGS, "Collation has all readings" );
is( scalar $c->paths, $PATHS, "Collation has all paths" );
is( scalar $c->relationships, 0, "Collation has all relationships" );

# Add a few relationships
$c->add_relationship( 'w123', 'w125', { 'type' => 'collated' } );
$c->add_relationship( 'w193', 'w196', { 'type' => 'collated' } );
$c->add_relationship( 'w257', 'w262', { 'type' => 'transposition' } );

# Now write it to GraphML and parse it again.

my $graphml = $c->as_graphml;
my $st = Text::Tradition->new( 'input' => 'Self', 'string' => $graphml );
is( scalar $st->collation->readings, $READINGS, "Reparsed collation has all readings" );
is( scalar $st->collation->paths, $PATHS, "Reparsed collation has all paths" );
is( scalar $st->collation->relationships, 3, "Reparsed collation has new relationships" );

# Now add a stemma, write to GraphML, and look at the output.
SKIP: {
	skip "Analysis module not present", 3 unless $tradition->can( 'add_stemma' );
	my $stemma = $tradition->add_stemma( 'dotfile' => 't/data/florilegium.dot' );
	is( ref( $stemma ), 'Text::Tradition::Stemma', "Parsed dotfile into stemma" );
	is( $tradition->stemmata, 1, "Tradition now has the stemma" );
	$graphml = $c->as_graphml;
	like( $graphml, qr/digraph/, "Digraph declaration exists in GraphML" );
}
}


# =begin testing
{
use Text::Tradition;
use Text::CSV;

my $READINGS = 311;
my $PATHS = 361;
my $WITS = 13;
my $WITAC = 4;

my $datafile = 't/data/florilegium_tei_ps.xml';
my $tradition = Text::Tradition->new( 'input' => 'TEI',
                                      'name' => 'test0',
                                      'file' => $datafile,
                                      'linear' => 1 );

my $c = $tradition->collation;
# Export the thing to CSV
my $csvstr = $c->as_csv();
# Count the columns
my $csv = Text::CSV->new({ sep_char => ',', binary => 1 });
my @lines = split(/\n/, $csvstr );
ok( $csv->parse( $lines[0] ), "Successfully parsed first line of CSV" );
is( scalar( $csv->fields ), $WITS + $WITAC, "CSV has correct number of witness columns" );
my $t2 = Text::Tradition->new( input => 'Tabular',
							   name => 'test2',
							   string => $csvstr,
							   sep_char => ',' );
is( scalar $t2->collation->readings, $READINGS, "Reparsed CSV collation has all readings" );
is( scalar $t2->collation->paths, $PATHS, "Reparsed CSV collation has all paths" );

# Now do it with TSV
my $tsvstr = $c->as_tsv();
my $t3 = Text::Tradition->new( input => 'Tabular',
							   name => 'test3',
							   string => $tsvstr,
							   sep_char => "\t" );
is( scalar $t3->collation->readings, $READINGS, "Reparsed TSV collation has all readings" );
is( scalar $t3->collation->paths, $PATHS, "Reparsed TSV collation has all paths" );

my $noaccsv = $c->as_csv({ noac => 1 });
my @noaclines = split(/\n/, $noaccsv );
ok( $csv->parse( $noaclines[0] ), "Successfully parsed first line of no-ac CSV" );
is( scalar( $csv->fields ), $WITS, "CSV has correct number of witness columns" );
}


# =begin testing
{
use Text::Tradition;

my $cxfile = 't/data/Collatex-16.xml';
my $t = Text::Tradition->new( 
    'name'  => 'inline', 
    'input' => 'CollateX',
    'file'  => $cxfile,
    );
my $c = $t->collation;

# Make an svg
my $table = $c->alignment_table;
ok( $c->has_cached_table, "Alignment table was cached" );
is( $c->alignment_table, $table, "Cached table returned upon second call" );
$c->calculate_ranks;
is( $c->alignment_table, $table, "Cached table retained with no rank change" );
$c->add_relationship( 'n13', 'n23', { type => 'repetition' } );
is( $c->alignment_table, $table, "Alignment table unchanged after non-colo relationship add" );
$c->add_relationship( 'n24', 'n23', { type => 'spelling' } );
isnt( $c->alignment_table, $table, "Alignment table changed after colo relationship add" );
}


# =begin testing
{
use Text::Tradition;

my $cxfile = 't/data/Collatex-16.xml';
my $t = Text::Tradition->new( 
    'name'  => 'inline', 
    'input' => 'CollateX',
    'file'  => $cxfile,
    );
my $c = $t->collation;

my @common = $c->calculate_common_readings();
is( scalar @common, 8, "Found correct number of common readings" );
my @marked = sort $c->common_readings();
is( scalar @common, 8, "All common readings got marked as such" );
my @expected = qw/ n1 n11 n16 n19 n20 n5 n6 n7 /;
is_deeply( \@marked, \@expected, "Found correct list of common readings" );
}


# =begin testing
{
use Text::Tradition;

my $cxfile = 't/data/Collatex-16.xml';
my $t = Text::Tradition->new( 
    'name'  => 'inline', 
    'input' => 'CollateX',
    'file'  => $cxfile,
    );
my $c = $t->collation;

is( $c->common_predecessor( 'n24', 'n23' )->id, 
    'n20', "Found correct common predecessor" );
is( $c->common_successor( 'n24', 'n23' )->id, 
    '__END__', "Found correct common successor" );

is( $c->common_predecessor( 'n19', 'n17' )->id, 
    'n16', "Found correct common predecessor for readings on same path" );
is( $c->common_successor( 'n21', 'n10' )->id, 
    '__END__', "Found correct common successor for readings on same path" );
}


1;
Commit	Line	Data
0e47f4f6	1	#!/usr/bin/perl -w
	2
	3	use strict;
	4	use Test::More 'no_plan';
	5	$\| = 1;
	6
	7
	8
	9	# =begin testing
	10	{
	11	use Text::Tradition;
	12
4e483aa5	13	my $cxfile = 't/data/Collatex-16.xml';
	14	my $t = Text::Tradition->new(
	15	'name' => 'inline',
	16	'input' => 'CollateX',
	17	'file' => $cxfile,
	18	);
	19	my $c = $t->collation;
	20
	21	my $rno = scalar $c->readings;
3c234eb6	22	# Split n21 ('unto') for testing purposes
4e483aa5	23	my $new_r = $c->add_reading( { 'id' => 'n21p0', 'text' => 'un', 'join_next' => 1 } );
	24	my $old_r = $c->reading( 'n21' );
	25	$old_r->alter_text( 'to' );
	26	$c->del_path( 'n20', 'n21', 'A' );
	27	$c->add_path( 'n20', 'n21p0', 'A' );
	28	$c->add_path( 'n21p0', 'n21', 'A' );
7a0956c1	29	$c->add_relationship( 'n21', 'n22', { type => 'collated', scope => 'local' } );
4e483aa5	30	$c->flatten_ranks();
	31	ok( $c->reading( 'n21p0' ), "New reading exists" );
	32	is( scalar $c->readings, $rno, "Reading add offset by flatten_ranks" );
	33
679f17e1	34	# Combine n3 and n4 ( with his )
4e483aa5	35	$c->merge_readings( 'n3', 'n4', 1 );
	36	ok( !$c->reading('n4'), "Reading n4 is gone" );
	37	is( $c->reading('n3')->text, 'with his', "Reading n3 has both words" );
	38
679f17e1	39	# Collapse n9 and n10 ( rood / root )
	40	$c->merge_readings( 'n9', 'n10' );
	41	ok( !$c->reading('n10'), "Reading n10 is gone" );
	42	is( $c->reading('n9')->text, 'rood', "Reading n9 has an unchanged word" );
4e483aa5	43
	44	# Combine n21 and n21p0
	45	my $remaining = $c->reading('n21');
	46	$remaining \|\|= $c->reading('n22'); # one of these should still exist
	47	$c->merge_readings( 'n21p0', $remaining, 1 );
	48	ok( !$c->reading('n21'), "Reading $remaining is gone" );
	49	is( $c->reading('n21p0')->text, 'unto', "Reading n21p0 merged correctly" );
	50	}
	51
	52
	53
	54	# =begin testing
	55	{
	56	use Text::Tradition;
f97ef19e	57
	58	my $st = Text::Tradition->new( 'input' => 'Self', 'file' => 't/data/collatecorr.xml' );
	59	is( ref( $st ), 'Text::Tradition', "Got a tradition from test file" );
	60	ok( $st->has_witness('Ba96'), "Tradition has the affected witness" );
	61
	62	my $sc = $st->collation;
	63	my $numr = 17;
	64	ok( $sc->reading('n131'), "Tradition has the affected reading" );
	65	is( scalar( $sc->readings ), $numr, "There are $numr readings in the graph" );
	66	is( $sc->end->rank, 14, "There are fourteen ranks in the graph" );
	67
	68	# Detach the erroneously collated reading
2dcb5d11	69	my( $newr, @del_rdgs ) = $sc->duplicate_reading( 'n131', 'Ba96' );
ef73c20a	70	ok( $newr, "New reading was created" );
f97ef19e	71	ok( $sc->reading('n131_0'), "Detached the bad collation with a new reading" );
	72	is( scalar( $sc->readings ), $numr + 1, "A reading was added to the graph" );
	73	is( $sc->end->rank, 10, "There are now only ten ranks in the graph" );
3c234eb6	74	my $csucc = $sc->common_successor( 'n131', 'n131_0' );
3c234eb6	75	is( $csucc->id, 'n136', "Found correct common successor to duped reading" );
f97ef19e	76
f97ef19e	77	# Check that the bad transposition is gone
2dcb5d11	78	is( scalar @del_rdgs, 1, "Deleted reading was returned by API call" );
f97ef19e	79	is( $sc->get_relationship( 'n130', 'n135' ), undef, "Bad transposition relationship is gone" );
f97ef19e	80
e19635f8	81	# The collation should not be fixed
	82	my @pairs = $sc->identical_readings();
	83	is( scalar @pairs, 0, "Not re-collated yet" );
f97ef19e	84	# Fix the collation
3c234eb6	85	ok( $sc->merge_readings( 'n124', 'n131_0' ), "Collated the readings correctly" );
e19635f8	86	@pairs = $sc->identical_readings( start => 'n124', end => $csucc->id );
3c234eb6	87	is( scalar @pairs, 3, "Found three more identical readings" );
e19635f8	88	is( $sc->end->rank, 11, "The ranks shifted appropriately" );
3c234eb6	89	$sc->flatten_ranks();
f97ef19e	90	is( scalar( $sc->readings ), $numr - 3, "Now we are collated correctly" );
	91	}
	92
	93
	94
	95	# =begin testing
	96	{
	97	use Text::Tradition;
951ddfe8	98	use TryCatch;
4e483aa5	99
56eefa04	100	my $READINGS = 311;
	101	my $PATHS = 361;
	102
	103	my $datafile = 't/data/florilegium_tei_ps.xml';
	104	my $tradition = Text::Tradition->new( 'input' => 'TEI',
	105	'name' => 'test0',
	106	'file' => $datafile,
	107	'linear' => 1 );
	108
	109	ok( $tradition, "Got a tradition object" );
	110	is( scalar $tradition->witnesses, 13, "Found all witnesses" );
	111	ok( $tradition->collation, "Tradition has a collation" );
	112
	113	my $c = $tradition->collation;
	114	is( scalar $c->readings, $READINGS, "Collation has all readings" );
	115	is( scalar $c->paths, $PATHS, "Collation has all paths" );
	116	is( scalar $c->relationships, 0, "Collation has all relationships" );
	117
	118	# Add a few relationships
	119	$c->add_relationship( 'w123', 'w125', { 'type' => 'collated' } );
	120	$c->add_relationship( 'w193', 'w196', { 'type' => 'collated' } );
	121	$c->add_relationship( 'w257', 'w262', { 'type' => 'transposition' } );
	122
	123	# Now write it to GraphML and parse it again.
	124
	125	my $graphml = $c->as_graphml;
	126	my $st = Text::Tradition->new( 'input' => 'Self', 'string' => $graphml );
	127	is( scalar $st->collation->readings, $READINGS, "Reparsed collation has all readings" );
	128	is( scalar $st->collation->paths, $PATHS, "Reparsed collation has all paths" );
	129	is( scalar $st->collation->relationships, 3, "Reparsed collation has new relationships" );
2a812726	130
9fef629b	131	# Now add a stemma, write to GraphML, and look at the output.
951ddfe8	132	SKIP: {
37bf09f4	133	skip "Analysis module not present", 3 unless $tradition->can( 'add_stemma' );
951ddfe8	134	my $stemma = $tradition->add_stemma( 'dotfile' => 't/data/florilegium.dot' );
	135	is( ref( $stemma ), 'Text::Tradition::Stemma', "Parsed dotfile into stemma" );
	136	is( $tradition->stemmata, 1, "Tradition now has the stemma" );
	137	$graphml = $c->as_graphml;
	138	like( $graphml, qr/digraph/, "Digraph declaration exists in GraphML" );
	139	}
56eefa04	140	}
	141
	142
	143
	144	# =begin testing
	145	{
16203db5	146	use Text::Tradition;
34ca808b	147	use Text::CSV;
16203db5	148
	149	my $READINGS = 311;
	150	my $PATHS = 361;
34ca808b	151	my $WITS = 13;
34ca808b	152	my $WITAC = 4;
16203db5	153
	154	my $datafile = 't/data/florilegium_tei_ps.xml';
	155	my $tradition = Text::Tradition->new( 'input' => 'TEI',
	156	'name' => 'test0',
	157	'file' => $datafile,
	158	'linear' => 1 );
	159
	160	my $c = $tradition->collation;
	161	# Export the thing to CSV
	162	my $csvstr = $c->as_csv();
34ca808b	163	# Count the columns
	164	my $csv = Text::CSV->new({ sep_char => ',', binary => 1 });
	165	my @lines = split(/\n/, $csvstr );
	166	ok( $csv->parse( $lines[0] ), "Successfully parsed first line of CSV" );
	167	is( scalar( $csv->fields ), $WITS + $WITAC, "CSV has correct number of witness columns" );
16203db5	168	my $t2 = Text::Tradition->new( input => 'Tabular',
	169	name => 'test2',
	170	string => $csvstr,
	171	sep_char => ',' );
	172	is( scalar $t2->collation->readings, $READINGS, "Reparsed CSV collation has all readings" );
	173	is( scalar $t2->collation->paths, $PATHS, "Reparsed CSV collation has all paths" );
	174
	175	# Now do it with TSV
	176	my $tsvstr = $c->as_tsv();
	177	my $t3 = Text::Tradition->new( input => 'Tabular',
	178	name => 'test3',
	179	string => $tsvstr,
	180	sep_char => "\t" );
	181	is( scalar $t3->collation->readings, $READINGS, "Reparsed TSV collation has all readings" );
	182	is( scalar $t3->collation->paths, $PATHS, "Reparsed TSV collation has all paths" );
34ca808b	183
	184	my $noaccsv = $c->as_csv({ noac => 1 });
	185	my @noaclines = split(/\n/, $noaccsv );
	186	ok( $csv->parse( $noaclines[0] ), "Successfully parsed first line of no-ac CSV" );
	187	is( scalar( $csv->fields ), $WITS, "CSV has correct number of witness columns" );
16203db5	188	}
	189
	190
	191
	192	# =begin testing
	193	{
56eefa04	194	use Text::Tradition;
56eefa04	195
0e47f4f6	196	my $cxfile = 't/data/Collatex-16.xml';
	197	my $t = Text::Tradition->new(
	198	'name' => 'inline',
	199	'input' => 'CollateX',
	200	'file' => $cxfile,
	201	);
	202	my $c = $t->collation;
4633f9e4	203
b365fbae	204	# Make an svg
bfcbcecb	205	my $table = $c->alignment_table;
	206	ok( $c->has_cached_table, "Alignment table was cached" );
	207	is( $c->alignment_table, $table, "Cached table returned upon second call" );
b365fbae	208	$c->calculate_ranks;
bfcbcecb	209	is( $c->alignment_table, $table, "Cached table retained with no rank change" );
864ee4bf	210	$c->add_relationship( 'n13', 'n23', { type => 'repetition' } );
	211	is( $c->alignment_table, $table, "Alignment table unchanged after non-colo relationship add" );
	212	$c->add_relationship( 'n24', 'n23', { type => 'spelling' } );
	213	isnt( $c->alignment_table, $table, "Alignment table changed after colo relationship add" );
b365fbae	214	}
	215
	216
	217
	218	# =begin testing
	219	{
	220	use Text::Tradition;
	221
	222	my $cxfile = 't/data/Collatex-16.xml';
	223	my $t = Text::Tradition->new(
	224	'name' => 'inline',
	225	'input' => 'CollateX',
	226	'file' => $cxfile,
	227	);
	228	my $c = $t->collation;
0e47f4f6	229
d4b75f44	230	my @common = $c->calculate_common_readings();
	231	is( scalar @common, 8, "Found correct number of common readings" );
	232	my @marked = sort $c->common_readings();
	233	is( scalar @common, 8, "All common readings got marked as such" );
679f17e1	234	my @expected = qw/ n1 n11 n16 n19 n20 n5 n6 n7 /;
d4b75f44	235	is_deeply( \@marked, \@expected, "Found correct list of common readings" );
	236	}
	237
	238
	239
	240	# =begin testing
	241	{
	242	use Text::Tradition;
	243
	244	my $cxfile = 't/data/Collatex-16.xml';
	245	my $t = Text::Tradition->new(
	246	'name' => 'inline',
	247	'input' => 'CollateX',
	248	'file' => $cxfile,
	249	);
	250	my $c = $t->collation;
	251
679f17e1	252	is( $c->common_predecessor( 'n24', 'n23' )->id,
0e47f4f6	253	'n20', "Found correct common predecessor" );
679f17e1	254	is( $c->common_successor( 'n24', 'n23' )->id,
10e4b1ac	255	'__END__', "Found correct common successor" );
0e47f4f6	256
4e5a7b2c	257	is( $c->common_predecessor( 'n19', 'n17' )->id,
0e47f4f6	258	'n16', "Found correct common predecessor for readings on same path" );
679f17e1	259	is( $c->common_successor( 'n21', 'n10' )->id,
10e4b1ac	260	'__END__', "Found correct common successor for readings on same path" );
0e47f4f6	261	}
	262
	263
	264
	265
	266	1;