Text-mode chess in C++ for the OOP2 final. Plays two-human or human-vs-AI.

API docs are live at https://caissa-docs.ufazien.com (generated from the Doxygen comments under src/).

Building

You need g++ (any version with C++17 — 6.3 works) and GNU Make.

make           # both binaries
make chess     # core only, what graders run
make chess-ai  # core + AI
make clean

The Makefile detects the shell, so it works from cmd.exe, PowerShell, Git Bash, and Linux/macOS terminals. Binaries land in bin/.

Running

./bin/chess                       # two-human game
./bin/chess-ai                    # asks "Play vs AI?" at startup
cat tests/games/4-mate/scholars.txt | ./bin/chess

Tests

make test

Runs the course's tests/test-level.sh against bin/chess for every level (1 → 4). The reference games live under tests/data/.

There's also make format (clang-format) and make doc (Doxygen) if you want them.

Notes on the design

The board owns every piece via std::unique_ptr<Piece>, so there is no manual delete anywhere. Captured pieces aren't actually removed from the roster — they keep a captured_ flag instead. This means execute_move and undo_move are reversible without any allocation, which is what makes the alpha-beta search cheap enough to feel instant at depth 3.

Every legal-move enumeration (mate detection, stalemate detection, engine move generation) goes through one templated helper, Board::for_each_legal_move. Single source of truth for what "legal" means, so I can't accidentally have the engine and the rules disagree.

---

Ufazien Deployment

The Doxygen-generated API docs are hosted on Ufazien at https://caissa-docs.ufazien.com.

Build and deploy

1. Generate the docs locally:

make doc            # writes HTML into docs/api/

2. Push to Ufazien:

ufazien deploy      # uploads docs/api/