commit 0af9909126309ca5da68e93761acb0cb6166398b
author Marc Kupietz <kupietz@ids-mannheim.de> Sat Feb 11 11:45:15 2023 +0100
committer Akron <nils@diewald-online.de> Tue Feb 14 11:00:02 2023 +0100
tree fa1e8938f0daa430cb07a422378abe050cff39d9
parent 997aa220641f64d75ba4c115083f7f8d5ae6695c

Set max heap size of KorAP tokenizer to 512m

Change-Id: I463d3f1f16a8267b3f6e79d076923ce2326631fc

lib/KorAP/XML/TEI/Tokenizer/KorAP.pm

diff --git a/lib/KorAP/XML/TEI/Tokenizer/KorAP.pm b/lib/KorAP/XML/TEI/Tokenizer/KorAP.pm
index 2061382..f38ba6c 100644
--- a/lib/KorAP/XML/TEI/Tokenizer/KorAP.pm
+++ b/lib/KorAP/XML/TEI/Tokenizer/KorAP.pm
@@ -33,7 +33,7 @@
 # Construct a new KorAP Tokenizer
 sub new {
   my ($class, $sentence_split) = @_;
-  my $self = $class->SUPER::new("$java -jar $tokenizer_jar --no-tokens --positions" .
+  my $self = $class->SUPER::new("$java -Xmx512m -jar $tokenizer_jar --no-tokens --positions" .
       ($sentence_split? " --sentence-boundaries" : ""));
   $self->{sentence_split} = $sentence_split;
   $self->{name} = 'korap';