Skip to content

Latest commit

 

History

History
115 lines (93 loc) · 6.43 KB

DEVNOTE.md

File metadata and controls

115 lines (93 loc) · 6.43 KB
Raw

UTBot Developer Guide

How to install docker image

UTBot has many dependencies, so the easiest way to build the tool from source and develop it is to use the docker container.

UTBot has a published docker package called base_env. It contains all the needed dependencies such as Git, LLVM, GRPC, GoogleTest and others. base_env has multiple versions tagged with dates. If you are developing the tool, you are most likely to need the most recent version from here.

Supported and tested development configuration are Ubuntu 18.04 / 20.04 or Windows Subsystem for Linux (Ubuntu 18.04 / 20.04).

  1. Install docker for Ubuntu or WSL+Ubuntu

  2. Install docker plugin that allows to mount host filesystem and remap owner and group:

    docker plugin install lebokus/bindfs

  3. Login into github docker registry:

    docker login -u <github-username> -p <github-personal-access-token> ghcr.io

    You can create new on this github page. Don't forget to check read:packages permission.
    More details are described here.

  4. Run docker/utbot_docker_dev.sh script. It will unpack the docker image and mount UTBot sources inside it. UTBot binary can be built in the docker and run in it. The script will prompt you to enter docker image tag. You can find the most recent tag here, for example 24-11-2021.

    Installer will ask about ssh_port on host machine. This port will be forwarded inside container's ssh port (by default sshd in container listens 2020). You may then login inside docker via ssh utbot@host -p $ssh_port, where host is a host machine IP address.

    ⚠ Specify free port that is not in use on host machine!

    Also, you will be prompted to enter a gRPC port on host machine that will be forwarded inside container's 2121 port where UTBot listens gRPC requests.

    Script will run docker image, mount specified folder on host filesystem into container's filesystem and forward ports for ssh and gRPC.

  5. Login via ssh into newly started container

    ssh utbot@host -p $ssh_port

    You can type localhost as host if you are inside terminal of a host machine) If you are prompted a password, enter utbot.

Install UTBotCpp server from source

  1. Clone UTBotCpp repository into home directory inside docker container, preferably with ssh.
  2. cd into UTBotCpp directory and run build.sh — it is the script that builds KLEE UTBot and runs UTBot unit tests
  3. Clone submodules git submodule update --init --recursive
  4. Navigate to UTBotCpp/server/build directory and launch the binary with ./utbot server command. Now the server is running.

How to develop UTBotCpp with VS Code

  1. Launch VS Code on your local machine. Use VS Code Remote-SSH to get access to the docker insides. Navigate to UTBotCpp/server directory and run build.sh script.
  2. Install necessary plugins C/C++ Extension Pack , GDB Debug
  3. Add debug configuration Run -> Add configuration and choose (gdb) launch set:
    • "program": "${workspaceFolder}/build/utbot"
    • "args": ["server", "--verbosity=trace"]
  4. Configure from CMake
  5. Build and Run

How to develop UTBotCpp with an CLion

You can edit UTBot sources, rebuild and rerun the server.

To setup it:

  1. Open UTBotCpp project in CLion on your local machine.
  2. Open Settings > Build, Execution, Deployment > Toolchains.
  3. Add a new Remote Host toolchain.
  4. In Credentials section, set up an ssh connection to the UTBot docker container you have. Consider enabling connection via OpehSSH authentification agent to ease files synchronization.
    • Cmake: /utbot_distr/install/bin/cmake
    • Makes: /usr/bin/make
    • C Compiler: /utbot_distr/install/bin/clang
    • C++ Compiler: /utbot_distr/install/bin/clang++
  6. Open Settings > Build, Execution, Deployment > Deployment in mappings Add new mapping from /utbot_distr/gtest to new local folder
  7. Sync files
  8. If you work on Windows, in docker install dos2unix and run find . -type f -print0 | xargs -0 dos2unix in UTBotCpp directory
  9. In docker run build.sh from project directory
  10. Load cmake from server/CMakeLists.txt

Build VS Code plugin

  1. Launch VS Code on your local machine. Use VS Code Remote-SSH to get access to the docker insides. Navigate to UTBotCpp/vscode-plugin directory and run build.sh script.
  2. Press F5 (Run Extension). This will run UTBot VS Code plugin.
  3. A new VS Code window will open; this window will have UTBot VS Code plugin enabled. In this new folder, open UTBotCpp/integration-tests/c-example directory.
  4. When UTBot Quickstart Wizard requests you to enter server host and port, specify localhost and UTBot server run port 2121, respectively. Oops, something went wrong! Please look at wizardInstall.gif
  5. Select project path as /home/utbot/UTBotCpp/integration-tests/c-example;
  6. You are now ready to experience UTBot capabilities! You can view possible commands in Command Palette (Press F1 and type in UTBot).

If you want to change UTBot test generation preferences, you can edit them in File > Preferences > Settings > Extensions > UnitTestBot. After UTBot configuration, you can select your source directories with the tab on the VSCode toolbar on the left. Then, you can generate tests with the use of Command Palette. Press F1 and type in "UTBot": You will see tests generation options.

Troubleshooting