From d20910fb8a63bbda30d4214d79884f5b697d8fe1 Mon Sep 17 00:00:00 2001
From: Dawid Jurkiewicz <dawid.jurkiewicz@applica.ai>
Date: Fri, 3 Apr 2020 14:00:11 +0200
Subject: [PATCH] metric stats accumulation

---
 Dockerfile      | 1 +
 Jenkinsfile     | 1 +
 calc_metrics.sh | 8 ++++++--
 3 files changed, 8 insertions(+), 2 deletions(-)

diff --git a/Dockerfile b/Dockerfile
index 758ce2d..7850706 100644
--- a/Dockerfile
+++ b/Dockerfile
@@ -5,6 +5,7 @@ RUN apt install -y git
 RUN apt install -y gcc
 RUN gcc --version
 RUN apt install -y build-essential
+RUN apt install -y moreutils
 RUN python --version
 RUN git clone https://github.com/usnistgov/SCTK.git
 WORKDIR SCTK
diff --git a/Jenkinsfile b/Jenkinsfile
index ecb4ff7..0a3471b 100644
--- a/Jenkinsfile
+++ b/Jenkinsfile
@@ -5,6 +5,7 @@ pipeline {
             steps {
                 checkout([$class: 'GitSCM', branches: [[name: '*/master']], doGenerateSubmoduleConfigurations: false, extensions: [], submoduleCfg: [], userRemoteConfigs: [[credentialsId: '8dbda16d-2863-4ee8-8662-3886a4862575', url: 'https://git.wmi.amu.edu.pl/dawjur/dawjur-mlworkshops']]])
                 copyArtifacts filter: 'wikiniews_results.tsv', fingerprintArtifacts: true, projectName: 'ASR-eval', selector: lastSuccessful()
+                copyArtifacts excludes: 'wikinews_results.tsv', fingerprintArtifacts: true, projectName: 'dawjur-metrics', selector: lastSuccessful()
             }
         }
         stage('Count lines') {
diff --git a/calc_metrics.sh b/calc_metrics.sh
index 6c3aa6b..c951269 100755
--- a/calc_metrics.sh
+++ b/calc_metrics.sh
@@ -3,9 +3,13 @@ cut -f2 wikiniews_results.tsv | awk 'BEGIN{FS=OFS="\t"}{print $0,"(sp1_"NR")"}'
 cut -f3 wikiniews_results.tsv | awk 'BEGIN{FS=OFS="\t"}{print $0,"(sp1_"NR")"}' > ref.trn
 sclite -f 0 -r ref.trn trn -h hyp.trn trn -e utf-8 -i rm -o sum stdout | grep "Sum/Avg" | python -c "import sys; import re; print('\t'.join(re.findall(r'(\d+\.\d+)', sys.stdin.read())[-2:]))" > wer_srr.txt
 cut -f1 wer_srr.txt > wer_total.txt
-cut -f2 wer_srr.txt > srr.txt
+cut -f2 wer_srr.txt >> srr.txt
 
 sclite -f 0 -r ref.trn trn -h hyp.trn trn -e utf-8 -i rm -o pra stdout | grep "Scores:" | sed 's/Scores: (#C #S #D #I) // ' | awk '{ c=$1; s=$2; d=$3; i=$4; print ($s+$d+$i)/($s+$d+$c)}' > wer_all.txt
-awk '{ sum += $1; n++ } END { print sum / n; }' < wer_all.txt > wer.txt
+awk '{ sum += $1; n++ } END { print sum / n; }' < wer_all.txt >> wer.txt
+
 paste wikiniews_results.tsv wer.txt > wikinews_results.tsv
 rm wikiniews_results.tsv
+
+tail -n20 wer.txt | sponge wer.txt
+tail -n20 srr.txt | sponge srr.txt