Hi, I am very excited about this project! 👍

I git clone master branch and opened the Playground but I get a compile error on the

import Sage

Line, with the error "No such module 'Sage'".

Any ideas?

First of all, awesome work you have here!

Now, how do I get the captured piece after a move?

I've been looking at the move history, but I wonder if you though about an easier way.

I don't know whether it is deliberate for not adding the code self.enPassantTarget = game.enPassantTarget. But not adding this line will cause the following issue:

Let's suppose a game A is in an "En Passant" state, that is, the next player who moves the piece can make an en passant move. Apparently, calling myGame.isLegal(move: myEnPassantMove) will return true. But if we call myGame.copy().isLegal(move: myEnPassantMove), it will return false which is not act as "expected".

i found that Minimax with alpha beta pruning is best way to implement chess AI, but i just wonder if i could using some other open source that could do this ?

The castling rights don't seem to be updated when a rook is captured. Here's a quick test I put together:

func testCastlingRightsAfterRookCapture() {
    let startFen = "rnbqkbnr/pppppp1p/8/8/1P4p1/8/PBPPPPPP/RN1QKBNR w KQkq - 0 1"
    let startPosition = Game.Position(fen: startFen)!
    let game = try! Game(position: startPosition)
    let move = Move(start: .b2, end: .h8)
    try! game.execute(move: move)
    XCTAssertFalse(game.castlingRights.contains(.blackKingside))
}

It looks like _execute(uncheckedMove:promotion:) is only updating castling rights when moving rooks or kings right now.

Interestingly, I was able to generate a fen that included invalid castling rights but I couldn't initialize a Position using that same fen (a PositionError.missingRook was thrown). It might be good to include position validation during fen generation too.

Thanks for the fantastic library. Very impressive work!

CocoaPods and Carthage are awesome tools and make our life really easier, but there are some devs who still don't know how to use them.

It would be cool to add the Manual installation guide in your README.md. You can take a look at my iOS Readme Template to see how you can do it.

nvzqz/Sage now has a Chat Room on Gitter

@nvzqz has just created a chat room. You can visit it here: https://gitter.im/nvzqz/Sage.

This pull-request adds this badge to your README.md:

If my aim is a little off, please let me know.

Happy chatting.

PS: Click here if you would prefer not to receive automatic pull-requests from Gitter in future.

Swift 4.2 introduced random seeding for hashing so that hash values are really different each time the program runs. The board initializer function in line 360 of Board.swift uses hash values as identifier for an array with 12 elements which under Swift 4.2/XCode 10 causes an index out of range error:

/// Create a chess board.
///
/// - parameter variant: The variant to populate the board for. Won't populate if `nil`. Default is `Standard`.
public init(variant: Variant? = ._standard) {
    #if swift(>=3)
        _bitboards = Array(repeating: 0, count: 12)
    #else
        _bitboards = Array(count: 12, repeatedValue: 0)
    #endif
    if let variant = variant {
        for piece in Piece.all {
            _bitboards[piece.hashValue] = Bitboard(startFor: piece)
        }
        if variant.isUpsideDown {
            for index in _bitboards.indices {
                _bitboards[index].flipVertically()
            }
        }
    }
}

The critical code is in the for loop: _bitboards[piece.hashValue] = Bitboard(startFor: piece) now crashes the program. I understand why the error is there but don't know why this program ran without error under Swift 4.1 because the hash value should (also under Swift 4.1) be a kind of random Int which may be bigger than the maximum index number 11 of the _bitboards array. @nvzqz seemed to have used the piece.hashValue rather like an identifier or rawValue.

Any help in explaining why the program didn't crash under Swift 4.1 and in rectifying the error would be appreciated!

When your own king is in check and your move removes the check and at the same time delivers check to the opponent king then the game.kingIsChecked variable is wrong, it is false but should be true.

I stumbled upon the error with the following position FEN: 2K1r3/3P1k2/8/8/8/8/8/2R5 w - - 0 1 where White is in check already. After the promotion move dxe8Q+ the black king is in check but kingIsChecked is false. After the promotion move d8N+ the black king is also in check but kingIsChecked is false.

The error has nothing to do with the promotion, I tried it with different positions and the common condition for the error is that the own king is in check directly before you put the enemy king in check.

En Passant move

An inconsistency may be created by performing undoMove() and redoMove() several times.

For example: Let a game starts with the following moves: e2->e4, f7->f5, e4->f5. Then, let's perform undoMove and redoMove. Then check the available moves for black pawn in g7. An inconsistency may be expected.

Reason: In private function _redoMove(), we calls _execute(uncheckedMove: promotion:) and passes the move that will be redone. But the property enPassantMove has not been updated in _redoMove. So there will be an inconsistency in the future calls performed on the game.

Solution: Updates the enPassantMove in _redoMove().

King is checked

kingIsChecked may returns the wrong value.

Reason: This is caused by copying a wrong value to _undoHistory in the function _undoMove(). Solution: Pass the correct value to _undoHistory.append(:).