ci is a command line interface for building and testing competitive programming tasks. Currently, it can test any executable and build C++ code.

cargo install --git https://github.com/matcegla/ci

If you do not have cargo installed, run curl https://sh.rustup.rs | sh.

ci generate-autocomplete bash | sudo tee /usr/share/bash-completion/completions/ci

ci build kitties.cpp

Compiles kitties.cpp to ./kitties.e. The code is built by compiler clang++ with C++17(-std=c++17) and warnings enabled(-Wall -Wextra -Wconversion -Wno-sign-conversion). This is the debug build, which includes debugging info(-g) and enables C++ standard library debug configuration (-D_GLIBCXX_DEBUG) as well as UB sanitizer(-fno-sanitize-recover=undefined), which means that you can debug it and also it will crash on UB.

ci build --release kitties.cpp

Compiles kitties.c to ./kitties.e. The --release flag changes the debug build to release build, which enables optimisations(-O2).

ci build --standard 11

Compiles with C++11 instead of default(C++17), with the flag -std=c++11. Possible values are: 11(-std=c++11), 17(-std=c++17).

ci test ./kitties.e kitties.test/

Recursively finds every file with .in extension in kitties.test/ directory, runs ./kitties.e with it as input, and compares if result is same as a matching file with .out extensions(ignoring some whitespace).

ci test ./kitties.e kitties.test/ --no-print-success

As before, but doesn't print information about tests that succeded. Useful when you have a lot of tests.

ci test ./kitties.e kitties.test/ --checker ./kitties-checker.e

Instead of comparing the output more or less char-by-char, uses a supplied program to do the checking. Program will be called like ./kitties-checker.e kitties.test/1.in <(./kitties.e < ./kitties.test/1.in) kitties.test/1.out and should return 0 status code if solution is valid(this will change).

ci multitest ./kitties-gen.py ./kitties-brut.e ./kitties.e ./kitties-alternative.e

Instead of using already generated tests, this uses ./kitties-gen.py to generate the inputs and runs them against every executable. First executable is assumed to always produce correct output.

ci multitest ./kitties-gen.py ./kitties-brut.e ./kitties.e ./kitties-alternative.e --checker ./kitties-checker.e

Instead of comparing char-by-char minus whitespace, this uses a checker like --checker in ci test does.

ci multitest ./kitties-gen.py ./kitties-brut.e ./kitties.e ./kitties-alternative.e -n 200

Instead of breaking on first failing test, this will run 200 tests and only print one with greatest fitness. Fitness function is - (test char count) by default.

ci multitest ./kitties-gen.py ./kitties-brut.e ./kitties.e -n 200 --fitness ./kitties-fit.e

Uses a supplied fitness function instead of the default. @bytelen is a special value that is -(test char count).

ci vendor kitties.cpp

When running your code in testing environments, you will not have access to various helpful header-only libraries like your personal algorithm collections, mc or Boost.Graph. This command will only run the preprocessor, copy-pasting these libraries into your code so you can send it to an online judge system.

To use the command, a directory /usr/share/ci/dummy-includes must be created, with all the includes you do not want copy-pasted(like <iostream>). To do this, create a file in this directory, like sudo touch /usr/share/ci/dummy-includes/iostream.

ci init https://codeforces.com/contest/960/problem/D

To save time on entering example tests, this command will download and save them to ./tests/example directory for you. Few task formats are supported.

ci submit code.cpp https://codeforces.com/contest/960/problem/D

To save time on submitting your code, this command will submit them for you! Few sites are suppported. The URL should be the URL to problem description(same as in ci init).