From 76e3ced286743988e2635eb489c40ce2ec06b35e Mon Sep 17 00:00:00 2001
From: Andrzej Preibisz <andrzejjpreibisz@gmail.com>
Date: Sun, 27 Mar 2022 13:28:05 +0200
Subject: [PATCH] Jenkinsfile + script fixes

---
 Jenkinsfile       | 14 ++++++++------
 dataset_script.sh | 14 +++++++++++---
 2 files changed, 19 insertions(+), 9 deletions(-)

diff --git a/Jenkinsfile b/Jenkinsfile
index acef5e1..6a60eeb 100644
--- a/Jenkinsfile
+++ b/Jenkinsfile
@@ -7,19 +7,21 @@ node {
                     description: 'Tekst do wyświetlenie',
                     name: 'INPUT_TEXT',
                     trim: false
+                ),
+                string(
+                    defaultValue: "10000",
+                    description: "How many records use",
+                    name: "CUTOFF"
                 )
             ])
         ])
     }
-    stage('Hello') {
-        //Wypisz wartość parametru w konsoli (To nie jest polecenie bash, tylko groovy!)
-        echo "INPUT_TEXT: ${INPUT_TEXT}"
-        //Wywołaj w konsoli komendę "figlet", która generuje ASCI-art
-        sh "figlet \"${INPUT_TEXT}\" | tee output.txt"
+    stage('Download') {
+        sh "dataset_script.sh \"${CUTOFF}\""
     }
     stage('Goodbye') {
         echo 'Goodbye!'
         //Zarchiwizuj wynik
-        archiveArtifacts 'output.txt'
+        archiveArtifacts 'heart_2020_test.csv, heart_2020_train.csv, heart_2020_sorted.csv'
     }
 }
\ No newline at end of file
diff --git a/dataset_script.sh b/dataset_script.sh
index f5776af..ef01464 100644
--- a/dataset_script.sh
+++ b/dataset_script.sh
@@ -1,3 +1,11 @@
+#!/bin/bash
+CUTOFF=$1
+
+TEST_SIZE=$(($CUTOFF * 1/5))
+TRAIN_START=$(($TEST_SIZE+1))
+echo $TEST_SIZE
+echo $TRAIN_START
+
 ! kaggle datasets download kamilpytlak/personal-key-indicators-of-heart-disease
 ! unzip personal-key-indicators-of-heart-disease -d ./
 ! wc -l heart_2020_cleaned.csv
@@ -9,10 +17,10 @@ echo "Categories: \n"
 
 
 ! grep -P "^$" -n heart_2020_cleaned.csv
-! head -n -1 heart_2020_cleaned.csv | shuf > heart_2020_sorted.csv
+! head -n $CUTOFF heart_2020_cleaned.csv | shuf > heart_2020_sorted.csv
 
-! head -n 4000 heart_2020_sorted.csv > heart_2020_test.csv
-! tail -n +4001 heart_2020_sorted.csv > heart_2020_train.csv
+! head -n $TEST_SIZE heart_2020_sorted.csv > heart_2020_test.csv
+! tail -n +$TRAIN_START heart_2020_sorted.csv > heart_2020_train.csv
 
 ! wc -l heart_2020_t*