From 941ab014266a23a7faf7710f38f86fb0f14f4879 Mon Sep 17 00:00:00 2001
From: Fabio Di Fabio <fabio.difabio@consensys.net>
Date: Tue, 16 Apr 2024 11:17:41 +0200
Subject: [PATCH] Split acceptance tests by time (#6953)

Signed-off-by: Fabio Di Fabio <fabio.difabio@consensys.net>
---
 .github/workflows/acceptance-tests.yml | 21 +++++------
 .github/workflows/splitTestsByTime.sh  | 52 ++++++++++++++++++++++++++
 2 files changed, 62 insertions(+), 11 deletions(-)
 create mode 100755 .github/workflows/splitTestsByTime.sh

diff --git a/.github/workflows/acceptance-tests.yml b/.github/workflows/acceptance-tests.yml
index 74ba75c2f9..7b212cdcad 100644
--- a/.github/workflows/acceptance-tests.yml
+++ b/.github/workflows/acceptance-tests.yml
@@ -35,6 +35,8 @@ jobs:
         with:
           distribution: temurin
           java-version: 17
+      - name: Install required packages
+        run:  sudo apt-get install -y xmlstarlet
       - name: get acceptance test report
         uses: dawidd6/action-download-artifact@e7466d1a7587ed14867642c2ca74b5bcc1e19a2d
         with:
@@ -48,19 +50,16 @@ jobs:
         with:
           cache-disabled: true
       - name: Split tests
-        id: split-tests
-        uses: r7kamura/split-tests-by-timings@9322bd292d9423e2bc5a65bec548901801341e3f
+        run: .github/workflows/splitTestsByTime.sh tmp/junit-xml-reports-downloaded ${{env.total-runners}} ${{ matrix.runner_index }} > testList.txt
+      - name: Upload Timing
+        uses: actions/upload-artifact@5d5d22a31266ced268874388b861e4b58bb5c2f3
+        if: matrix.runner_index == 0
         with:
-          reports: tmp/junit-xml-reports-downloaded
-          glob: 'acceptance-tests/tests/src/test/java/org/hyperledger/besu/tests/acceptance/**/*Test.java'
-          total: ${{env.total-runners}}
-          index: ${{ matrix.runner_index }}
-      - name: write out test list
-        run: echo "${{ steps.split-tests.outputs.paths }}" >> testList.txt
+          name: acceptance-tests-timing
+          path: 'tmp/timing.tsv'
       - name: format gradle args
-        #regex means: first truncate file paths to align with package name, then swap path delimiter with package delimiter,
-        #then drop file extension, then insert --tests option between each.
-        run: cat testList.txt | sed -e 's@acceptance-tests/tests/src/test/java/@--tests\ @g;s@/@.@g;s/\.java//g'  > gradleArgs.txt
+        # insert --tests option between each.
+        run:  cat testList.txt | sed -e 's/^\| / --tests /g' | tee gradleArgs.txt
       - name: run acceptance tests
         run: ./gradlew acceptanceTestNotPrivacy `cat gradleArgs.txt` -Dorg.gradle.parallel=true -Dorg.gradle.caching=true
       - name: cleanup tempfiles
diff --git a/.github/workflows/splitTestsByTime.sh b/.github/workflows/splitTestsByTime.sh
new file mode 100755
index 0000000000..3afb614d42
--- /dev/null
+++ b/.github/workflows/splitTestsByTime.sh
@@ -0,0 +1,52 @@
+#!/bin/bash
+
+REPORTS_DIR="$1"
+SPLIT_COUNT=$2
+SPLIT_INDEX=$3
+
+# extract tests time from Junit XML reports
+find "$REPORTS_DIR" -type f -name TEST-*.xml | xargs -I{} bash -c "xmlstarlet sel -t -v 'sum(//testcase/@time)' '{}'; echo '{}' | sed 's/.*TEST\-\(.*\)\.xml/ \1/'" > tmp/timing.tsv
+
+# Sort times in descending order
+IFS=$'\n' sorted=($(sort -nr tmp/timing.tsv))
+unset IFS
+
+sums=()
+tests=()
+
+# Initialize sums
+for ((i=0; i<SPLIT_COUNT; i++))
+do
+	sums[$i]=0
+done
+
+# add tests to groups trying to balance the sum of execution time of each group
+for line in "${sorted[@]}"; do
+	line_parts=( $line )
+	test_time=${line_parts[0]//./} # convert to millis
+	test_time=${test_time##0} # remove leading zeros
+	test_name=${line_parts[1]}
+	
+	# Find index of min sum
+	idx_min_sum=0
+	min_sum=${sums[0]}
+	for ((i=0; i<SPLIT_COUNT; i++))
+	do		
+		if [[ ${sums[$i]} -lt $min_sum ]]
+		then
+			idx_min_sum=$i		
+			min_sum=${sums[$i]}
+		fi
+	done
+
+	# Add the test to the min sum list
+	min_sum_tests=${tests[$idx_min_sum]}
+	tests[$idx_min_sum]="${min_sum_tests}${test_name},"
+	
+	# Update the sums
+	((sums[idx_min_sum]+=test_time))
+
+done
+
+# return the requests index, without quotes to drop the last trailing space
+echo ${tests[$SPLIT_INDEX]//,/ }
\ No newline at end of file