COMPETITIVE PROGRAMMING CLI (C++)

What is this?

This project provides a set of tools for testing the correctness of solutions in coding competitions. Because scripts are all command-line interface (CLI), the project is independent of text editors/ IDEs. vscode and nvim are the recommended text editors.

This project is in the early stage of development, so there is a lot of room for improvement. Nevertheless, it is mature enough for casual coding competition.

Install

Requirements

1. Nix

Step by step installation

1. Clone the repo, run nix build, then run ./install.sh.
2. Create your workspace directory with sub directories:
   • task
   • archive
   • output
3. Create new task and solve problem

You can also run cpcli directly from the checkout without installing it:

nix run . -- --project-config=/path/to/project_config.toml --help

Workspace Configuration

Sample Configuration

This file should be named project_config.toml and placed directly at the top level of your workspace folder. root is optional; when omitted, cpcli uses the directory containing project_config.toml. ~ and relative paths are supported.

task_editor_exec = "cpcli_editor"
use_template_engine = false

[language_config]
default = "cpp"

[language_config.override]

[language_config.cpp]
compiler = "g++"
regular_flag = "-DLOCAL -O2 -std=c++17"
debug_flag = "-DLOCAL -Wall -Wshadow -std=c++17 -g -fsanitize=address -fsanitize=undefined -D_GLIBCXX_DEBUG"
use_precompiled_header = false
use_cache = true
include_dir = "include"

[language_config.cu]
compiler = "cpcli_cuda_compile"
runtime = "cpcli_cuda_run"
regular_flag = "-std=c++17 -O2 -lineinfo -Wno-deprecated-gpu-targets -gencode arch=compute_70,code=compute_70"
debug_flag = "-std=c++17 -O0 -g -lineinfo -Wno-deprecated-gpu-targets -gencode arch=compute_70,code=compute_70"

[language_config.py]
interpreter = "python3"

CUDA tasks use solution.cu. When runtime is set, cpcli compiles a solution.cuda-bin and creates ./solution as a launcher that passes the CUDA binary to that runtime. This allows simulator or container backends without changing the task runner.

The bundled cpcli_cuda_setup command extracts the matching CUDA headers and libdevice into ~/.cache/cpcli/cuda-toolkit-12.8 for clangd/editor support.

Sample folder structure

<your workspace>
├── archive
│   ├── Archive
│   ├── AtCoder - ACL Beginner Contest
│   ├── TopCoder SRM #456
│   ├── Topcoder - TCO 2021 Regional Qualifier 1 DIV 1
│   ├── Topcoder 2021 Round 1B
│   ├── Topcoder Open Algo 2019
│   ├── Unsorted
│   └── vnoi.info
├── include
│   ├── genlib.hpp
│   ├── interactive.hpp
│   └── testlib.h
├── output
│   └── solution.cpp
├── project_config.toml
├── task
│   └── F - Keep Connect
│       ├── config.toml
│       └── solution.cpp
└── template
    ├── checker.template
    ├── gen.template
    ├── interactor.template
    └── solution.template

Please take a look at the archive folder for more information.

Sample invocation

$ cpcli_app --project-config=project_config.toml task --root-dir="/Users/cunbidun/competitive_programming/task/F - Keep Connect" --build

Num	Attribute	Type	Description	Default value
1	task_dir	string	Directory for storing current task
2	output_dir	string	Compiled solution will be copy to this directory (so you know where to look for file for submission)	""
3	template_dir	string	Directory for storing template (check the repo template folder for more info)	""
4	archive_dir	string	Directory for archiving completed task (check the repo archive folder for more info)	""
6	include_dir	string	Store libs here
5	task_editor_exec	string	Executable frontend to edit task config.toml or config.json	cpcli_editor
7	cpp_compile_flag	string	Cpp normal complier flag	"-DLOCAL -static -O2 -std=c++17"
8	cpp_debug_flag	string	Cpp debug flag	"-DLOCAL -Wall -Wshadow -std=c++17 -g -fsanitize=address -fsanitize=undefined -D_GLIBCXX_DEBUG"
9	use_precompiled_header	bool	use precompiled headers	true
10	use_cache	bool	Enable cache	true

Task Configuration

New task configs are written as config.toml. Existing flat config.json and old flat TOML keys are still accepted, but saving through the task editor writes the new sectioned TOML shape.

tests = [
  { enabled = true, has_expected_output = true, input = "1\n", output = "1\n" },
]

[problem]
name = "E. Trees of Tranquillity"
group = "Codeforces - Codeforces Round #722 (Div. 2)"
url = "https://codeforces.com/contest/1529/problem/E"

[run]
time_limit_ms = 3000
checker = "custom"
interactive = false
stop_on_first_failure = false

[display]
hide_accepted_tests = false
truncate_long_output = false

[generation]
enabled = true
test_count = 10
seed = ""
args = ""
expected_output_from_slow = true

Canonical task config keys:

Attribute	Type	Description	Default value
problem.name	string	Problem name or id.	""
problem.group	string	Contest/group name used for archive.	""
problem.url	string	Problem URL.	unset
run.time_limit_ms	int	Time limit in milliseconds.	10000
run.checker	string	Checker name.	token_checker
run.interactive	boolean	Whether the task is interactive.	false
run.stop_on_first_failure	boolean	Stop after WA, RTE, or TLE.	false
display.hide_accepted_tests	boolean	Hide accepted test details.	true
display.truncate_long_output	boolean	Truncate long input/output when printing.	false
generation.enabled	boolean	Enable generated tests.	false
generation.test_count	int	Number of generated tests.	0
generation.seed	string	Generator seed. If empty, cpcli creates one.	""
generation.args	string	Extra arguments passed to gen after seed and count.	""
generation.expected_output_from_slow	boolean	Use slow to produce expected output for generated tests.	false
tests[].enabled	boolean	Whether this sample test runs.	true
tests[].has_expected_output	boolean	Whether output should be checked.	derived from output
tests[].input	string	Test input.	required
tests[].output	string	Expected output.	optional

Others checker included:

1. double_4: checking up to 4 decimals digit
2. double_6: checking up to 6 decimals digit
3. double_9: checking up to 9 decimals digit
4. token_checker: compare output files token by token
5. custom: user's custom checker

Development

Environment Setup

The recommend text editor for developing this project is vscode

1. Enter the development environment and configure the project:

   nix develop
   cmake -S . -B build -G Ninja -DBUILD_TESTING=ON

2. Build the project. CMake writes build/compile_commands.json for editor tooling.

   cmake --build build

3. Run the tests:

   ctest --test-dir build --output-on-failure

4. .vscode setup

   Create a .vscode at the top level directory of the repository

   .vscode
   ├── c_cpp_properties.json
   └── settings.json
   

   c_cpp_properties.json content:

   {
     "configurations": [
       {
         "name": "cpcli_app",
        "compileCommands": "${workspaceFolder}/build/compile_commands.json"
       }
     ],
     "version": 4
   }

Build and Run Task Editor

nix build
./install.sh
result/share/cpcli/task-editor/task-editor --root ~/competitive_programming/task/<task>

Build and run GTest

cmake -S . -B build -G Ninja -DBUILD_TESTING=ON
cmake --build build
ctest --test-dir build --output-on-failure

Build code coverage report

Build with coverage instrumentation, then run the tests and generate the report:

cmake -S . -B build-coverage -G Ninja -DBUILD_TESTING=ON \
  -DCMAKE_CXX_FLAGS="--coverage" -DCMAKE_EXE_LINKER_FLAGS="--coverage"
cmake --build build-coverage
ctest --test-dir build-coverage
lcov --capture --directory build-coverage --output-file coverage.info
genhtml coverage.info --output-directory coverage
open coverage/index.html

Build Documentations

Requirements

Make sure to have sphinx and sphinx-rtd-theme installed

$ pip install -U sphinx sphinx-rtd-theme

Build and run docs

$ cd docs
$ make clean html
$ open build/html/index.html

Credits

1. The project is heavily inspired by Egor Kulikov's idea-chelper.

2. Checkers and generators use Mike Mirzayanov's testlib

3. The precompiled header feature is inspired by Dushyant Singh's script Speed-Up-GCC-Compile-Time.