CodeCharta also comes as a full Docker environment, where all supported tools come pre-installed. To start all containers, you can type docker compose pull && docker compose up --detach while you’re in the same directory as our docker-compose.yml

The Container Landscape

Container-Name Description How to use
sonar Hosts an instance of SonarQube localhost:9000 in the browser, follow the steps for Linux
codecharta-visualization Runs the visualisation part o CodeChara localhost:9001, load files from your hard-drive
codecharta-analysis Contains all tools the ccsh can import Connect via terminal docker exec -it codecharta-analysis bash

To see the actual names of the containers on your system, run docker ps

All containers share a volume for the quick transfer of files. You can find it under /mnt/data in each container. Please note that you will need to copy finished cc.json files to your hard-drive to open them in Visualization.

Docker Hub Install

The visualization is published to Docker Hub.

Visualization

# run visualization with
docker run -p 80:8080 codecharta/codecharta-visualization

Sonar

See also SonarQube Docs

Open localhost:9000 in your browser and log in with

  • login: admin
  • password: admin

Simply follow the steps for a manual, local project under Linux. You can also check our tutorial for SonarQube The sonar-scanner is already pre-installed in our analysis container.

codecharta-visualization

See also CodeCharta Visualization

Open localhost:9001 in your browser and open any file you want from your hard drive. To open files you have created in the analysis container, copy them over using docker cp

codecharta-analysis

See also CodeCharta Analysis

Almost all tools the ccsh can import data from are included in the container, so you can get started immediately. Installed are:

  • CodeCharta Shell ccsh
  • Git git
  • Tokei tokei
  • CodeMaat java -jar /opt/codemaat/codemaat.jar
  • MetricGardener metric-gardener
  • SonarScanner sonar-scanner

Once you have your .cc.json file ready, you can copy it to your machine. This is how the command could look like if I want to copy a file from the container to my current working directory:

docker cp codecharta-analysis:/root/junit4.cc.json.gz junit4.cc.json.gz

To check the name of the container, you can simply type docker ps. To analyze your code you can follow one of our quick-start guides or check out the in-depth documentation for a certain importer.

Hands-On

Example execution of codecharta-analysis from the compose project:

docker run -it -v $(pwd):/src -w /src codecharta/codecharta-analysis ccsh gitlogparser repo-scan --repo-path /src -o my-project.cc.json -nc

# Explanation
# -it = interactive, tty-Terminal: Connects your terminal to the execution of your command
# -v = virtual-mount: Mount your current working directory inside the container to /src
#               Hint: You need to mount a absolute path, $(pwd) will print your working directory
# -w = working-directory:  Starting point of your command, in this case the /src folder where the directory was mounted to
# codecharta/codecharta-analysis: tag of the image
# gitlogparser [...] : check the gitlogparser documentation for more info

Be aware, that by default the user inside the docker image is ‘ubuntu’ with an ID of 1000. You may encounter errors with git when you try to execute commands inside a repository cloned by a different UID

For macOS users, it is necessary to add --user=501:dialout Before the image name. Why? Because docker on mac is fun :)

Build it yourself

You can also build a docker image from the locally installed instance of codecharta. To do this, you would execute the following commands:

# to build the container you need to navigate into the analysis folder
cd analysis

# makes a clean build to ensure everything will be built correctly
./gradlew clean build

# builds the docker container with an image name of your choice
docker build . -t your-image-name

# executes bash inside of the created container
docker run -it your-image-name bash

# to run the docker container with your current working directory mounted into it,
# you can alternatively run this
docker run -it -v $(pwd):/src -w /src your-image-name bash

There is also the option to customize the used docker user during the build. This will create a new user with the given name and IDs.

#USER_ADD: set it to any value to to add a new user
#USERNAME: the name of the user to be created
#USER_ID: the ID that is used for the new user and for its group
docker build ./analysis -t local-ccsh --build-arg USER_ADD=true --build-arg USER_ID=1001 --build-arg USERNAME=your_name

Updated: