| #!/usr/bin/env perl |
| use strict; |
| use warnings; |
| |
| # Comparison path |
| my $cmd = '/euralex/corpus/empirist_gold_cmc/tools/compare_tokenization.perl'; |
| |
| # Output path |
| my $ud_path = '/euralex/ud_tokens'; |
| mkdir $ud_path; |
| |
| my $base = 'de_gsd-ud-train.conllu'; |
| |
| # Split files |
| chdir '/euralex/corpus/'; |
| system 'perl /euralex/benchmarks/cleanup/split_conllu.pl /euralex/corpus/' . $base; |
| chdir '/euralex'; |
| |
| my $gold = '/euralex/corpus/' . $base . '.split'; |
| my $raw = '/euralex/corpus/' . $base . '.raw'; |
| |
| my %tools = ( |
| waste => sub { |
| system 'cat ' . $raw . ' | waste -N -v0 --rcfile=./Waste/waste.rc > ' . $ud_path . '/waste/' . $base; |
| }, |
| datok => sub { |
| system 'cat ' . $raw . ' | ./Datok/datok tokenize -t ./Datok/testdata/tokenizer.matok - > ' . $ud_path . '/datok/' . $base; |
| }, |
| korap_tokenizer => sub { |
| system 'cat ' . $raw . ' | java -jar ./KorAP-Tokenizer/KorAP-Tokenizer.jar -l de > ' . $ud_path . '/korap_tokenizer/' . $base; |
| }, |
| opennlp_simple => sub { |
| system 'cat ' . $raw . ' | ./opennlp/bin/opennlp SimpleTokenizer 2> /dev/null | sed "s/\s/\n/g" > ' . $ud_path . '/opennlp_simple/' . $base; |
| }, |
| opennlp_tokenizer => sub { |
| system 'cat ' . $raw . ' | ./opennlp/bin/opennlp TokenizerME ./opennlp/models/opennlp-de-ud-gsd-tokens-1.0-1.9.3.bin 2> /dev/null | sed "s/\s/\n/g" > ' . $ud_path . '/opennlp_tokenizer/' . $base; |
| }, |
| tree_tagger => sub { |
| system 'cat ' . $raw . ' | perl ./treetagger/cmd/utf8-tokenize.perl -a ./treetagger/lib/german-abbreviations 2> /dev/null > ' . $ud_path . '/tree_tagger/' . $base; |
| }, |
| jtok => sub { |
| chdir '/euralex/JTok/bin'; |
| system 'sh tokenize ' . $raw . ' de | grep "Token: " | perl -CS -pe "s/\s +Token: \"//; s/^(\"?[^\"]*?)\".+?$/\1/g" > ' . $ud_path . '/jtok/' . $base; |
| chdir '/euralex'; |
| }, |
| syntok => sub { |
| system 'python3 -m syntok.tokenizer ' . $raw . ' | sed "s/\s/\n/g" > ' . $ud_path . '/syntok/' . $base; |
| }, |
| elephant => sub { |
| system './elephant-wrapper/bin/tokenize.sh -i ' . $raw . ' UD_German | sed "s/\s/\n/g" > ' . $ud_path . '/elephant/' . $base; |
| }, |
| spacy => sub { |
| system 'python3 ./spacy/spacy_tok.py ' . $raw . ' > ' . $ud_path . '/spacy/' . $base; |
| }, |
| cutter => sub { |
| system 'python3 ./cutter/cutter.py nosent ' . $raw . ' > ' . $ud_path . '/cutter/' . $base; |
| }, |
| somajo => sub { |
| system 'somajo-tokenizer ' . $raw . ' 2> /dev/null > ' . $ud_path . '/somajo/' . $base; |
| }, |
| stanford => sub { |
| system 'CLASSPATH=/euralex/stanford-corenlp-4.4.0/* java edu.stanford.nlp.pipeline.StanfordCoreNLP ' . |
| '-props german -annotators tokenize,ssplit,mwt -tokenize.language=german -file ' . $raw . ' 2> /dev/null'; |
| system 'perl /euralex/benchmarks/cleanup/stanford.pl ' . $base . '.raw.out > ' . $ud_path . '/stanford/' . $base; |
| system 'rm ' . $base . '.raw.out'; |
| } |
| ); |
| |
| # delete $tools{waste}; |
| # delete $tools{datok}; |
| # delete $tools{korap_tokenizer}; |
| # delete $tools{opennlp_simple}; |
| # delete $tools{opennlp_tokenizer}; |
| # delete $tools{tree_tagger}; |
| # delete $tools{jtok}; |
| # delete $tools{syntok}; |
| # delete $tools{somajo}; |
| # delete $tools{stanford}; |
| # delete $tools{elephant}; |
| # delete $tools{spacy}; |
| # delete $tools{cutter}; |
| |
| # Create project folders |
| foreach (keys %tools) { |
| mkdir $ud_path . '/' . $_; |
| }; |
| |
| # Run tokenization |
| foreach (keys %tools) { |
| $tools{$_}->(); |
| }; |
| |
| foreach my $tool (keys %tools) { |
| print "\n##########\n"; |
| print "##### $tool - UD\n"; |
| print "##\n"; |
| system $cmd . ' -x ' . $gold . ' ' . $ud_path . '/' . $tool . '/' . $base . ' 2> /dev/null'; |
| }; |