More fixes for meta data paths
diff --git a/lib/KorAP/Document.pm b/lib/KorAP/Document.pm
index 08aa2d0..4e648c3 100644
--- a/lib/KorAP/Document.pm
+++ b/lib/KorAP/Document.pm
@@ -10,7 +10,7 @@
use KorAP::Log;
use Mojo::DOM;
use Data::Dumper;
-use File::Spec::Functions qw/catdir catfile splitdir/;
+use File::Spec::Functions qw/catdir catfile catpath splitdir splitpath rel2abs/;
our @ATTR = qw/text_sigle
doc_sigle
@@ -57,8 +57,11 @@
sub new {
my $class = shift;
my $self = bless { @_ }, $class;
- if (exists $self->{path} && $self->{path} !~ m!\/$!) {
- $self->{path} .= '/';
+ if (exists $self->{path}) {
+ $self->{path} = rel2abs($self->{path});
+ if ($self->{path} !~ m!\/$!) {
+ $self->{path} .= '/';
+ };
};
return $self;
};
@@ -121,11 +124,11 @@
croak $unable;
};
- my @path = splitdir($self->path);
+ my @path = grep { $_ } splitdir($self->path);
my @header;
foreach (0..2) {
- unshift @header, catfile(@path, 'header.xml');
+ unshift @header, '/' . catfile(@path, 'header.xml');
pop @path;
};
my @type = qw/corpus doc text/;
diff --git a/t/real_goethe.t b/t/real_goethe.t
index 20addf1..14437f3 100644
--- a/t/real_goethe.t
+++ b/t/real_goethe.t
@@ -21,6 +21,7 @@
# GOE/AGA/03828
my $path = catdir(dirname(__FILE__), 'GOE/AGA/03828');
+# my $path = '/home/ndiewald/Repositories/korap/KorAP-sandbox/KorAP-lucene-indexer/t/GOE/AGA/03828';
ok(my $doc = KorAP::Document->new( path => $path . '/' ), 'Load Korap::Document');
ok($doc->parse, 'Parse document');
diff --git a/t/real_goethe_benchmark.t b/t/real_goethe_benchmark.t
index f119a2b..66152dd 100644
--- a/t/real_goethe_benchmark.t
+++ b/t/real_goethe_benchmark.t
@@ -4,6 +4,7 @@
use strict;
use warnings;
use Test::More;
+use Mojo::ByteStream 'b';
use Benchmark qw/:hireswallclock/;
@@ -19,6 +20,7 @@
# GOE/AGA/03828
my $path = catdir(dirname(__FILE__), 'GOE/AGA/03828');
+# Todo: Test with absolute path!
ok(my $doc = KorAP::Document->new( path => $path . '/' ), 'Load Korap::Document');
ok($doc->parse, 'Parse document');
@@ -50,18 +52,20 @@
ok($tokens->add('CoreNLP', 'Morpho'), 'Add corenlp morpho');
ok($tokens->add('CoreNLP', 'Constituency'), 'Add corenlp constituency');
ok($tokens->add('Glemm', 'Morpho'), 'Add glemm morpho');
-ok($tokens->add('Connexor', 'Sentences'), 'Add cnx sentences');
-ok($tokens->add('Connexor', 'Morpho'), 'Add cnx morpho');
-ok($tokens->add('Connexor', 'Phrase'), 'Add cnx phrase');
-ok($tokens->add('Connexor', 'Syntax'), 'Add cnx syntax');
+# t ok($tokens->add('Connexor', 'Sentences'), 'Add cnx sentences');
+# t ok($tokens->add('Connexor', 'Morpho'), 'Add cnx morpho');
+# t ok($tokens->add('Connexor', 'Phrase'), 'Add cnx phrase');
+# t ok($tokens->add('Connexor', 'Syntax'), 'Add cnx syntax');
ok($tokens->add('Mate', 'Morpho'), 'Add mate morpho');
# $tokens->add('Mate', 'Dependency');
-ok($tokens->add('XIP', 'Sentences'), 'Add xip sentences');
-ok($tokens->add('XIP', 'Morpho'), 'Add xip morpho');
-ok($tokens->add('XIP', 'Constituency'), 'Add xip constituency');
+# t ok($tokens->add('XIP', 'Sentences'), 'Add xip sentences');
+# t ok($tokens->add('XIP', 'Morpho'), 'Add xip morpho');
+# t ok($tokens->add('XIP', 'Constituency'), 'Add xip constituency');
# $tokens->add('XIP', 'Dependency');
ok($tokens->to_json, 'To json');
is($tokens->doc->to_hash->{title}, 'Autobiographische Einzelheiten');
+b($tokens->to_json)->spurt('AGA.03828.json');
+
diag timestr(timediff(Benchmark->new, $t));