How to integrate JaCoCo into a spring boot OCI image to measure E2E testing code coverage?

I found the Tanzu Jacoco Build Pack. Is this the right way?

If you're a Tanzu/ Broadcom customer, maybe, otherwise it's not an OSS component, so no.

And it seems like Paketo Buildpacks do not have a Jacoco buildpack, so...let's rather consider a workaround.

Running JaCoCo in "production" is just a matter of attaching the JaCoCo agent to the running JVM, something along the lines of:

$ java -javaagent:jacocoagent.jar=address=localhost,includes="*",port=36320,destfile=jacoco.exec,output=tcpserver -jar build/libs/my-app-0.0.1-SNAPSHOT.jar

You already know how to pass options to the JVM for the runtime, so now you need to figure out how to ship the agent and retrieve the jacoco.exec file.

The easiest way would be to mount a volume (or several) to your Spring Boot app at runtime, a read/write volume where your jacocoagent.jarwould sit and where you jacoco.exec would be written to.

Hope that makes sense, if you're having issues, come and ask the Paketo Community on Slack

Wonderful. That helped me very much.

I have now a separate "playwright" module with this build.gradle.kts:

plugins {
    base
    java
    jacoco
}

repositories {
    mavenCentral()
}
dependencies {
    runtimeOnly("org.jacoco:org.jacoco.agent:0.8.14:runtime")
}

val jacocoDir = layout.buildDirectory.dir("logs/jacoco")

tasks.named("check") {
    dependsOn("playwright")
    // report is always generated after tests run
    finalizedBy(tasks.jacocoTestReport)
}

tasks.register<Exec>("playwright") {
    dependsOn(":backend:bootBuildImage")
    dependsOn("copyJacocoLib")
    inputs.dir("./")
    commandLine("./run-docker.sh")
}

tasks.register<Copy>("copyJacocoLib") {
    from(configurations.runtimeClasspath)
    into(jacocoDir)
    include("org.jacoco.agent-*-runtime.jar")
}

jacoco {
    toolVersion = "0.8.14"
    reportsDirectory = jacocoDir
}

tasks.jacocoTestReport {
    executionData = files("${jacocoDir.get()}/jacoco.exec")
    sourceSets(project(":kicktipp-backend").sourceSets.main.get())
    reports {
        html.required = true
        csv.required = true
        xml.required = true
    }
}

And a compose file similar to this:

services:
  app:
    image: backend:dev
    volumes:
      - .${JACOCO_DIR}:${JACOCO_DIR}
    environment:
      - JAVA_TOOL_OPTIONS=-javaagent:${JACOCO_LIB}=destfile=${JACOCO_REPORT}
    healthcheck:
      test: [ "CMD", "curl", "-f", "http://localhost:8080/" ]
      start_period: 10s
      start_interval: 1s
      retries: 10

  playwright:
    image: mcr.microsoft.com/playwright:v1.52.0-noble
    # https://playwright.dev/docs/docker#recommended-docker-configuration
    init: true
    ipc: host
    user: ubuntu:ubuntu
    volumes:
      - ../:/app
    working_dir: /app/kicktipp-playwright
    command: bash -c "npm ci && npx playwright test"
    depends_on:
      app:
        condition: service_healthy

an .env file:

JACOCO_DIR=/build/logs/jacoco
JACOCO_VERSION=0.8.14
JACOCO_LIB=${JACOCO_DIR}/org.jacoco.agent-${JACOCO_VERSION}-runtime.jar
JACOCO_REPORT=${JACOCO_DIR}/jacoco.exec

And my docker-run.sh

#!/bin/bash

cd $(dirname $0)
set -e
LOG_DIR="build/logs"
LOG_FILE="$LOG_DIR/docker.log"
CONTAINER="playwright-playwright-1"
mkdir -p $LOG_DIR $LOG_DIR/playwright
trap '{
  docker logs --tail 100 $CONTAINER
}' ERR
docker compose --ansi never up --build --exit-code-from playwright >$LOG_FILE 2>&1

And of course a usual playwright configuration.

Not perfect yet, but it works and shows the coverage in gitlab:

build:
  stage: build
  script:
    - ./gradlew build
    - awk -F"," '{ instructions += $4 + $5; covered += $5 } END { print covered, "/", instructions, " instructions covered"; print 100*covered/instructions, "% covered" }' playwright/build/logs/jacoco/test/jacocoTestReport.csv
  coverage: /([0-9]{1,3}.[0-9]*).%.covered/

Better ways to do it but just for further reference if someone needs a code example to get it running

Hey Janning Vygen , I'm a Product Manager at Stack Overflow. We're gathering feedback on the new question type you used to ask this. If you have a moment, we'd love to hear how your experience was or what you would improve.

Havn't thought about it really. But I guess I like the question/anwser model more maybe because i am used to it.

But anwsering on your quesion was quite difficult, had to scroll to the top and click "Join the conersation".