Improve benchmarks and exclude from distro
Change-Id: If70fa19418ccfd339695997944ac34350742319a
diff --git a/MANIFEST b/MANIFEST
index 64a5c6f..e3d83d4 100644
--- a/MANIFEST
+++ b/MANIFEST
@@ -191,4 +191,3 @@
t/tokens.t
t/tools/example_rules.json
t/transform.t
-xt/benchmark.pl
diff --git a/MANIFEST.SKIP b/MANIFEST.SKIP
index f8efe15..c2a6eb1 100755
--- a/MANIFEST.SKIP
+++ b/MANIFEST.SKIP
@@ -1,5 +1,6 @@
# Skip real corpus regression files and tests
^t/real/
+^xt/
# Makemaker generated files and dirs.
^MANIFEST\.
diff --git a/xt/benchmark.pl b/xt/benchmark.pl
index bc470d5..69fc908 100644
--- a/xt/benchmark.pl
+++ b/xt/benchmark.pl
@@ -38,6 +38,16 @@
my $cache = tmpnam();
# end instance 1
+
+# begin instance 2 setup
+# Load example file
+use KorAP::XML::Krill;
+use KorAP::XML::Tokenizer;
+my $path = catdir(dirname(__FILE__), '..','t','real', 'corpus','GOE-TAGGED','AGA','03828');
+my ($tokens, $result);
+# end instance 2
+
+
# Create a new benchmark object
my $bench = Dumbbench->new(
verbosity => 0
@@ -61,6 +71,46 @@
);
`$cmd`;
}
+ ),
+ Dumbbench::Instance::PerlSub->new(
+ name => 'Conversion-GOE-Tagged-1',
+ code => sub {
+ my $doc = KorAP::XML::Krill->new(path => $path . '/');
+ $doc->parse;
+ my $meta = $doc->meta;
+ $tokens = KorAP::XML::Tokenizer->new(
+ path => $doc->path,
+ doc => $doc,
+ foundry => 'Base',
+ layer => 'Tokens_conservative',
+ name => 'tokens'
+ );
+ $tokens->parse;
+ $tokens->add('DeReKo', 'Structure', 'base_sentences_paragraphs');
+ $tokens->add('DRuKoLa', 'Morpho');
+ $result = $tokens->to_data;
+ $tokens = undef;
+ }
+ ),
+ Dumbbench::Instance::PerlSub->new(
+ name => 'Conversion-GOE-Tagged-1',
+ code => sub {
+ my $doc = KorAP::XML::Krill->new(path => $path . '/');
+ $doc->parse;
+ my $meta = $doc->meta;
+ $tokens = KorAP::XML::Tokenizer->new(
+ path => $doc->path,
+ doc => $doc,
+ foundry => 'Base',
+ layer => 'Tokens_conservative',
+ name => 'tokens'
+ );
+ $tokens->parse;
+ $tokens->add('DeReKo', 'Structure', 'base_sentences_paragraphs');
+ $tokens->add('DRuKoLa', 'Morpho');
+ $result = $tokens->to_data;
+ $tokens = undef;
+ }
)
);
@@ -76,6 +126,8 @@
exit(0);
};
+print "----------------------------------\n";
+
# Output simple timings for comparation
foreach my $inst ($bench->instances) {
unless ($no_header) {
diff --git a/xt/prof-conv-goe-tagged-1.pl b/xt/prof-conv-goe-tagged-1.pl
new file mode 100644
index 0000000..9ba0356
--- /dev/null
+++ b/xt/prof-conv-goe-tagged-1.pl
@@ -0,0 +1,28 @@
+#!/usr/bin/env perl
+use strict;
+use warnings;
+use File::Basename 'dirname';
+use File::Spec::Functions qw/catfile catdir rel2abs/;
+
+# Run
+# $ perl -d:NYTProf xt/prof-conv-goe-tagged-1.pl
+# $ nytprofhtml --open
+
+use KorAP::XML::Krill;
+use KorAP::XML::Tokenizer;
+my $path = catdir(dirname(__FILE__), '..','t','real', 'corpus','GOE-TAGGED','AGA','03828');
+
+my $doc = KorAP::XML::Krill->new(path => $path . '/');
+$doc->parse;
+my $meta = $doc->meta;
+my $tokens = KorAP::XML::Tokenizer->new(
+ path => $doc->path,
+ doc => $doc,
+ foundry => 'Base',
+ layer => 'Tokens_conservative',
+ name => 'tokens'
+);
+$tokens->parse;
+$tokens->add('DeReKo', 'Structure', 'base_sentences_paragraphs');
+$tokens->add('DRuKoLa', 'Morpho');
+$tokens->to_data;