python-chess: a pure Python chess library

https://travis-ci.org/niklasf/python-chess.svg?branch=master https://coveralls.io/repos/niklasf/python-chess/badge.png https://landscape.io/github/niklasf/python-chess/master/landscape.png

Introduction

This is the scholars mate in python-chess:

>>> import chess

>>> board = chess.Board()

>>> board.push_san("e4")
Move.from_uci('e2e4')
>>> board.push_san("e5")
Move.from_uci('e7e5')
>>> board.push_san("Qh5")
Move.from_uci('d1h5')
>>> board.push_san("Nc6")
Move.from_uci('b8c6')
>>> board.push_san("Bc4")
Move.from_uci('f1c4')
>>> board.push_san("Nf6")
Move.from_uci('g8f6')
>>> board.push_san("Qxf7")
Move.from_uci('h5f7')

>>> board.is_checkmate()
True

Features

  • Supports Python 2.7 and Python 3.

    >>> # Python 2 compability for the following examples.
    >>> from __future__ import print_function
    
  • Legal move generator and move validation. This includes all castling rules and en-passant captures.

    >>> chess.Move.from_uci("a8a1") in board.legal_moves
    False
    
  • Make and unmake moves.

    >>> Qf7 = board.pop() # Unmake last move (Qf7#)
    >>> Qf7
    Move.from_uci('h5f7')
    
    >>> board.push(Qf7) # Restore
    
  • Show a simple ASCII board.

    >>> print(board)
    r . b q k b . r
    p p p p . Q p p
    . . n . . n . .
    . . . . p . . .
    . . B . P . . .
    . . . . . . . .
    P P P P . P P P
    R N B . K . N R
    
  • Detects checkmates, stalemates and draws by insufficient material.

    >>> board.is_stalemate()
    False
    >>> board.is_insufficient_material()
    False
    >>> board.is_game_over()
    True
    >>> board.halfmove_clock
    0
    
  • Detects repetitions. Has a half move clock.

    >>> board.can_claim_threefold_repetition()
    False
    >>> board.halfmove_clock
    0
    >>> board.can_claim_fifty_moves()
    False
    >>> board.can_claim_draw()
    False
    

    With the new rules from July 2014 a game ends drawn (even without a claim) once a fivefold repetition occurs or if there are 75 moves without a pawn push or capture. Other ways of ending a game take precedence.

    >>> board.is_fivefold_repetition()
    False
    >>> board.is_seventyfive_moves()
    False
    
  • Detects checks and attacks.

    >>> board.is_check()
    True
    >>> board.is_attacked_by(chess.WHITE, chess.E8)
    True
    
    >>> attackers = board.attackers(chess.WHITE, chess.F3)
    >>> attackers
    SquareSet(0b100000001000000)
    >>> chess.G2 in attackers
    True
    
  • Parses and creates SAN representation of moves.

    >>> board = chess.Board()
    >>> board.san(chess.Move(chess.E2, chess.E4))
    'e4'
    
  • Parses and creates FENs.

    >>> board.fen()
    'rnbqkbnr/pppppppp/8/8/8/8/PPPPPPPP/RNBQKBNR w KQkq - 0 1'
    >>> board = chess.Board("8/8/8/2k5/4K3/8/8/8 w - - 4 45")
    >>> board.piece_at(chess.C5)
    Piece.from_symbol('k')
    
  • Parses and creates EPDs.

    >>> board = chess.Board()
    >>> board.epd(bm=chess.Move.from_uci("d2d4"))
    'rnbqkbnr/pppppppp/8/8/8/8/PPPPPPPP/RNBQKBNR w KQkq - bm d4;'
    
    >>> ops = board.set_epd("1k1r4/pp1b1R2/3q2pp/4p3/2B5/4Q3/PPP2B2/2K5 b - - bm Qd1+; id \"BK.01\";")
    >>> ops == {'bm': chess.Move.from_uci('d6d1'), 'id': 'BK.01'}
    True
    
  • Read Polyglot opening books.

    >>> import chess.polyglot
    
    >>> book = chess.polyglot.open_reader("data/opening-books/performance.bin")
    >>> board = chess.Board()
    >>> first_entry = next(book.get_entries_for_position(board))
    >>> first_entry.move()
    Move.from_uci('e2e4')
    >>> first_entry.learn
    0
    >>> first_entry.weight
    1
    
    >>> book.close()
    
  • Read and write PGNs. Supports headers, comments, NAGs and a tree of variations.

    >>> import chess.pgn
    
    >>> pgn = open("data/games/molinari-bordais-1979.pgn")
    >>> first_game = chess.pgn.read_game(pgn)
    >>> pgn.close()
    
    >>> first_game.headers["White"]
    'Molinari'
    >>> first_game.headers["Black"]
    'Bordais'
    
    >>> # Iterate through the mainline of this embarrasingly short game.
    >>> node = first_game
    >>> while node.variations:
    ...     next_node = node.variation(0)
    ...     print(node.board().san(next_node.move))
    ...     node = next_node
    e4
    c5
    c4
    Nc6
    Ne2
    Nf6
    Nbc3
    Nb4
    g3
    Nd3#
    
    >>> first_game.headers["Result"]
    '0-1'
    
  • Probe Syzygy endgame tablebases.

    >>> import chess.syzygy
    
    >>> tablebases = chess.syzygy.Tablebases("data/syzygy")
    
    >>> # Black to move is losing in 53 half moves (distance to zero) in this
    >>> # KNBvK endgame.
    >>> board = chess.Board("8/2K5/4B3/3N4/8/8/4k3/8 b - - 0 1")
    >>> tablebases.probe_dtz(board)
    -53
    
    >>> tablebases.close()
    
  • Communicate with an UCI engine.

    >>> import chess.uci
    >>> import time
    
    >>> engine = chess.uci.popen_engine("stockfish")
    >>> engine.uci()
    >>> engine.author
    'Tord Romstad, Marco Costalba and Joona Kiiski'
    
    >>> # Synchronous mode.
    >>> board = chess.Board("1k1r4/pp1b1R2/3q2pp/4p3/2B5/4Q3/PPP2B2/2K5 b - - 0 1")
    >>> engine.position(board)
    >>> engine.go(movetime=2000) # Gets tuple of bestmove and ponder move.
    BestMove(bestmove=Move.from_uci('d6d1'), ponder=Move.from_uci('c1d1'))
    
    >>> # Synchronous communication, but search in background.
    >>> engine.go(infinite=True)
    >>> time.sleep(2)
    >>> engine.stop()
    BestMove(bestmove=Move.from_uci('d6d1'), ponder=Move.from_uci('c1d1'))
    
    >>> # Asynchronous mode.
    >>> def callback(command):
    ...    bestmove, ponder = command.result()
    ...    assert bestmove == chess.Move.from_uci('d6d1')
    ...
    >>> command = engine.go(movetime=2000, async_callback=callback)
    >>> command.done()
    False
    >>> command.result()
    BestMove(bestmove=Move.from_uci('d6d1'), ponder=Move.from_uci('c1d1'))
    >>> command.done()
    True
    
    >>> # Quit.
    >>> engine.quit()
    0
    

Peformance

python-chess is not intended to be used by serious chess engines where performance is critical. The goal is rather to create a simple and relatively highlevel library.

You can install the gmpy2 or gmpy (https://code.google.com/p/gmpy/) modules in order to get a slight performance boost on basic operations like bit scans and population counts.

python-chess will only ever import very basic general (non-chess-related) operations from native libraries. All logic is pure Python. There will always be pure Python fallbacks.

Installing

  • With pip:

    sudo pip install python-chess
    
  • From current source code:

    python setup.py sdist
    sudo python setup.py install
    

License

python-chess is licensed under the GPL3. See the LICENSE file for the full copyright and license information.

Thanks to the developers of http://chessx.sourceforge.net/. Some of the core bitboard move generation parts are ported from there.

Thanks to Ronald de Man for his Syzygy endgame tablebases (https://github.com/syzygy1/tb). The probing code in python-chess is very directly ported from his C probing code.

Contents

Changelog for python-chess

This project is pretty young and maturing only slowly. At the current stage it is more important to get things right, than to be consistent with previous versions. Use this changelog to see what changed in a new release, because this might include API breaking changes.

New in v0.8.3

Bugfixes:

  • The initial move number in PGNs was missing, if black was to move in the starting position. Thanks to Jürgen Précour for reporting.
  • Detect more impossible en-passant squares in Board.status(). There already was a requirement for a pawn on the fifth rank. Now the sixth and seventh rank must be empty, additionally. We do not do further retrograde analysis, because these are the only cases affecting move generation.

New in v0.8.2

Bugfixes:

  • pgn.Game.setup() with the standard starting position was failing when the standard starting position was already set. Thanks to Jordan Bray for reporting this.

Optimizations:

  • Remove bswap() from Syzygy decompression hot path. Directly read integers with the correct endianness.

New in v0.8.1

  • Fixed pondering mode in uci module. For example ponderhit() was blocking indefinitely. Thanks to Valeriy Huz for reporting this.
  • Patch by Richard C. Gerkin: Moved searchmoves to the end of the UCI go command, where it will not cause other command parameters to be ignored.
  • Added missing check or checkmate suffix to castling SANs, e.g. O-O-O#.
  • Fixed off-by-one error in polyglot opening book binary search. This would not have caused problems for real opening books.
  • Fixed Python 3 support for reverse polyglot opening book iteration.
  • Bestmoves may be literally (none) in UCI protocol, for example in checkmate positions. Fix parser and return None as the bestmove in this case.
  • Fixed spelling of repetition (was repitition). can_claim_threefold_repetition() and is_fivefold_repetition() are the affected method names. Aliases are there for now, but will be removed in the next release. Thanks to Jimmy Patrick for reporting this.
  • Added SquareSet.__reversed__().
  • Use containerized tests on Travis CI, test against Stockfish 6, improved test coverage amd various minor clean-ups.

New in v0.8.0

  • Implement Syzygy endgame tablebase probing. https://syzygy-tables.info is an example project that provides a public API using the new features.
  • The interface for aynchronous UCI command has changed to mimic concurrent.futures. is_done() is now just done(). Callbacks will receive the command object as a single argument instead of the result. The result property and wait() have been removed in favor of a synchronously waiting result() method.
  • The result of the stop and go UCI commands are now named tuples (instead of just normal tuples).
  • Add alias Board for Bitboard.
  • Fixed race condition during UCI engine startup. Lines received during engine startup sometimes needed to be processed before the Engine object was fully initialized.

New in v0.7.0

  • Implement UCI engine communication.
  • Patch by Matthew Lai: Add caching for gameNode.board().

New in v0.6.0

  • If there are comments in a game before the first move, these are now assigned to Game.comment instead of Game.starting_comment. Game.starting_comment is ignored from now on. Game.starts_variation() is no longer true. The first child node of a game can no longer have a starting comment. It is possible to have a game with Game.comment set, that is otherwise completely empty.
  • Fix export of games with variations. Previously the moves were exported in an unusual (i.e. wrong) order.
  • Install gmpy2 or gmpy if you want to use slightly faster binary operations.
  • Ignore superfluous variation opening brackets in PGN files.
  • Add GameNode.san().
  • Remove sparse_pop_count(). Just use pop_count().
  • Remove next_bit(). Now use bit_scan().

New in v0.5.0

  • PGN parsing is now more robust: read_game() ignores invalid tokens. Still exceptions are going to be thrown on illegal or ambiguous moves, but this behaviour can be changed by passing an error_handler argument.

    >>> # Raises ValueError:
    >>> game = chess.pgn.read_game(file_with_illegal_moves)
    
    >>> # Silently ignores errors and continues parsing:
    >>> game = chess.pgn.read_game(file_with_illegal_moves, None)
    
    >>> # Logs the error, continues parsing:
    >>> game = chess.pgn.read_game(file_with_illegal_moves, logger.exception)
    

    If there are too many closing brackets this is now ignored.

    Castling moves like 0-0 (with zeros) are now accepted in PGNs. The Bitboard.parse_san() method remains strict as always, though.

    Previously the parser was strictly following the PGN spefification in that empty lines terminate a game. So a game like

    [Event "?"]
    
    { Starting comment block }
    
    1. e4 e5 2. Nf3 Nf6 *
    

    would have ended directly after the starting comment. To avoid this, the parser will now look ahead until it finds at least one move or a termination marker like *, 1-0, 1/2-1/2 or 0-1.

  • Introduce a new function scan_headers() to quickly scan a PGN file for headers without having to parse the full games.

  • Minor testcoverage improvements.

New in v0.4.2

  • Fix bug where pawn_moves_from() and consequently is_legal() weren’t handling en-passant correctly. Thanks to Norbert Naskov for reporting.

New in v0.4.1

  • Fix is_fivefold_repitition(): The new fivefold repitition rule requires the repititions to occur on alternating consecutive moves.
  • Minor testing related improvements: Close PGN files, allow running via setuptools.
  • Add recently introduced features to README.

New in v0.4.0

  • Introduce can_claim_draw(), can_claim_fifty_moves() and can_claim_threefold_repitition().
  • Since the first of July 2014 a game is also over (even without claim by one of the players) if there were 75 moves without a pawn move or capture or a fivefold repitition. Let is_game_over() respect that. Introduce is_seventyfive_moves() and is_fivefold_repitition(). Other means of ending a game take precedence.
  • Threefold repitition checking requires efficient hashing of positions to build the table. So performance improvements were needed there. The default polyglot compatible zobrist hashes are now built incrementally.
  • Fix low level rotation operations l90(), l45() and r45(). There was no problem in core because correct versions of the functions were inlined.
  • Fix equality and inequality operators for Bitboard, Move and Piece. Also make them robust against comparisons with incompatible types.
  • Provide equality and inequality operators for SquareSet and polyglot.Entry.
  • Fix return values of incremental arithmetical operations for SquareSet.
  • Make polyglot.Entry a collections.namedtuple.
  • Determine and improve test coverage.
  • Minor coding style fixes.

New in v0.3.1

  • Bitboard.status() now correctly detects STATUS_INVALID_EP_SQUARE, instead of errors or false reports.
  • Polyglot opening book reader now correctly handles knight underpromotions.
  • Minor coding style fixes, including removal of unused imports.

New in v0.3.0

  • Rename property half_moves of Bitboard to halfmove_clock.

  • Rename property ply of Bitboard to fullmove_number.

  • Let PGN parser handle symbols like !, ?, !? and so on by converting them to NAGs.

  • Add a human readable string representation for Bitboards.

    >>> print(chess.Bitboard())
    r n b q k b n r
    p p p p p p p p
    . . . . . . . .
    . . . . . . . .
    . . . . . . . .
    . . . . . . . .
    P P P P P P P P
    R N B Q K B N R
    
  • Various documentation improvements.

New in v0.2.0

  • Implement PGN parsing and writing.
  • Hugely improve test coverage and use Travis CI for continuous integration and testing.
  • Create an API documentation.
  • Improve Polyglot opening-book handling.

New in v0.1.0

Apply the lessons learned from the previous releases, redesign the API and implement it in pure Python.

New in v0.0.4

Implement the basics in C++ and provide bindings for Python. Obviously performance was a lot better - but at the expense of having to compile code for the target platform.

Pre v0.0.4

First experiments with a way too slow pure Python API, creating way too many objects for basic operations.

Core

Colors

Constants for the side to move or the color of a piece.

chess.WHITE = 0
chess.BLACK = 1

You can get the opposite color using color ^ 1.

Piece types

chess.NONE = 0
chess.PAWN
chess.KNIGHT
chess.BISHOP
chess.ROOK
chess.QUEEN
chess.KING

Castling rights

The castling flags

chess.CASTLING_NONE = 0
chess.CASTLING_WHITE_KINGSIDE
chess.CASTLING_BLACK_KINGSIDE
chess.CASTLING_WHITE_QUEENSIDE
chess.CASTLING_BLACK_QUEENSIDE

can be combined bitwise.

chess.CASTLING_WHITE = CASTLING_WHITE_QUEENSIDE | CASTLING_WHITE_KINGSIDE
chess.CASTLING_BLACK = CASTLING_BLACK_QUEENSIDE | CASTLING_BLACK_KINGSIDE
chess.CASTLING = CASTLING_WHITE | CASTLING_BLACK

Squares

chess.A1 = 0
chess.B1 = 1

and so on to

chess.H8 = 63
chess.SQUARES = [A1, B1, ..., G8, H8]
chess.SQUARE_NAMES = ['a1', 'b1', ..., 'g8', 'h8']
chess.file_index(square)

Gets the file index of square where 0 is the a file.

chess.FILE_NAMES = ['a', 'b', ..., 'g', 'h']
chess.rank_index(square)

Gets the rank index of the square where 0 is the first rank.

Pieces

class chess.Piece(piece_type, color)

A piece with type and color.

piece_type

The piece type.

color

The piece color.

symbol()

Gets the symbol P, N, B, R, Q or K for white pieces or the lower-case variants for the black pieces.

classmethod from_symbol(symbol)

Creates a piece instance from a piece symbol.

Raises ValueError if the symbol is invalid.

Moves

class chess.Move(from_square, to_square, promotion=0)

Represents a move from a square to a square and possibly the promotion piece type.

Castling moves are identified only by the movement of the king.

Null moves are supported.

from_square

The source square.

to_square

The target square.

promotion

The promotion piece type.

uci()

Gets an UCI string for the move.

For example a move from A7 to A8 would be a7a8 or a7a8q if it is a promotion to a queen. The UCI representatin of null moves is 0000.

classmethod from_uci(uci)

Parses an UCI string.

Raises ValueError if the UCI string is invalid.

classmethod null()

Gets a null move.

A null move just passes the turn to the other side (and possibly forfeits en-passant capturing). Null moves evaluate to False in boolean contexts.

>>> bool(chess.Move.null())
False

Board

chess.STARTING_FEN = 'rnbqkbnr/pppppppp/8/8/8/8/PPPPPPPP/RNBQKBNR w KQkq - 0 1'

The FEN notation of the standard chess starting position.

class chess.Board(fen=None)

A bitboard and additional information representing a position.

Provides move generation, validation, parsing, attack generation, game end detection, move counters and the capability to make and unmake moves.

The bitboard is initialized to the starting position, unless otherwise specified in the optional fen argument.

turn

The side to move.

castling_rights

Bitmask of castling rights.

ep_square

The potential en-passant square on the third or sixth rank or 0. It does not matter if en-passant would actually be possible on the next move.

fullmove_number

Counts move pairs. Starts at 1 and is incremented after every move of the black side.

halfmove_clock

The number of half moves since the last capture or pawn move.

A dynamic list of pseudo legal moves.

Pseudo legal moves might leave or put the king in check, but are otherwise valid. Null moves are not pseudo legal. Castling moves are only included if they are completely legal.

For performance moves are generated on the fly and only when nescessary. The following operations do not just generate everything but map to more efficient methods.

>>> len(board.pseudo_legal_moves)
20
>>> bool(board.pseudo_legal_moves)
True
>>> move in board.pseudo_legal_moves
True
legal_moves = LegalMoveGenerator(self)

A dynamic list of completely legal moves, much like the pseudo legal move list.

reset()

Restores the starting position.

clear()

Clears the board.

Resets move stacks and move counters. The side to move is white. There are no rooks or kings, so castling is not allowed.

In order to be in a valid status() at least kings need to be put on the board. This is required for move generation and validation to work properly.

piece_at(square)

Gets the piece at the given square.

piece_type_at(square)

Gets the piece type at the given square.

remove_piece_at(square)

Removes a piece from the given square if present.

set_piece_at(square, piece)

Sets a piece at the given square. An existing piece is replaced.

is_attacked_by(color, square)

Checks if the given side attacks the given square. Pinned pieces still count as attackers.

attackers(color, square)

Gets a set of attackers of the given color for the given square.

Returns a set of squares.

is_check()

Checks if the current side to move is in check.

is_into_check(move)

Checks if the given move would move would leave the king in check or put it into check.

was_into_check()

Checks if the king of the other side is attacked. Such a position is not valid and could only be reached by an illegal move.

is_game_over()

Checks if the game is over due to checkmate, stalemate, insufficient mating material, the seventyfive-move rule or fivefold repetition.

is_checkmate()

Checks if the current position is a checkmate.

is_stalemate()

Checks if the current position is a stalemate.

is_insufficient_material()

Checks for a draw due to insufficient mating material.

is_seventyfive_moves()

Since the first of July 2014 a game is automatically drawn (without a claim by one of the players) if the half move clock since a capture or pawn move is equal to or grather than 150. Other means to end a game take precedence.

is_fivefold_repetition()

Since the first of July 2014 a game is automatically drawn (without a claim by one of the players) if a position occurs for the fifth time on consecutive alternating moves.

is_fivefold_repitition()

Since the first of July 2014 a game is automatically drawn (without a claim by one of the players) if a position occurs for the fifth time on consecutive alternating moves.

can_claim_draw()

Checks if the side to move can claim a draw by the fifty-move rule or by threefold repetition.

can_claim_fifty_moves()

Draw by the fifty-move rule can be claimed once the clock of halfmoves since the last capture or pawn move becomes equal or greater to 100 and the side to move still has a legal move they can make.

can_claim_threefold_repetition()

Draw by threefold repetition can be claimed if the position on the board occured for the third time or if such a repetition is reached with one of the possible legal moves.

can_claim_threefold_repitition()

Draw by threefold repetition can be claimed if the position on the board occured for the third time or if such a repetition is reached with one of the possible legal moves.

push(move)

Updates the position with the given move and puts it onto a stack.

Null moves just increment the move counters, switch turns and forfeit en passant capturing.

No validation is performed. For performance moves are assumed to be at least pseudo legal. Otherwise there is no guarantee that the previous board state can be restored. To check it yourself you can use:

>>> move in board.pseudo_legal_moves
True
pop()

Restores the previous position and returns the last move from the stack.

peek()

Gets the last move from the move stack.

set_epd(epd)

Parses the given EPD string and uses it to set the position.

If present the hmvc and the fmvn are used to set the half move clock and the fullmove number. Otherwise 0 and 1 are used.

Returns a dictionary of parsed operations. Values can be strings, integers, floats or move objects.

Raises ValueError if the EPD string is invalid.

epd(**operations)

Gets an EPD representation of the current position.

EPD operations can be given as keyword arguments. Supported operands are strings, integers, floats and moves. All other operands are converted to strings.

hmvc and fmvc are not included by default. You can use:

>>> board.epd(hmvc=board.halfmove_clock, fmvc=board.fullmove_number)
'rnbqkbnr/pppppppp/8/8/8/8/PPPPPPPP/RNBQKBNR w KQkq - hmvc 0; fmvc 1;'
set_fen(fen)

Parses a FEN and sets the position from it.

Rasies ValueError if the FEN string is invalid.

fen()

Gets the FEN representation of the position.

parse_san(san)

Uses the current position as the context to parse a move in standard algebraic notation and return the corresponding move object.

The returned move is guaranteed to be either legal or a null move.

Raises ValueError if the SAN is invalid or ambigous.

push_san(san)

Parses a move in standard algebraic notation, makes the move and puts it on the the move stack.

Raises ValueError if neither legal nor a null move.

Returns the move.

san(move)

Gets the standard algebraic notation of the given move in the context of the current position.

There is no validation. It is only guaranteed to work if the move is legal or a null move.

status()

Gets a bitmask of possible problems with the position. Move making, generation and validation are only guaranteed to work on a completely valid board.

zobrist_hash(array=None)

Returns a Zobrist hash of the current position.

A zobrist hash is an exclusive or of pseudo random values picked from an array. Which values are picked is decided by features of the position, such as piece positions, castling rights and en-passant squares. For this implementation an array of 781 values is required.

The default behaviour is to use values from POLYGLOT_RANDOM_ARRAY, which makes for hashes compatible with polyglot opening books.

PGN parsing and writing

Game model

Games are represented as a tree of moves. Each GameNode can have extra information such as comments. The root node of a game (Game extends GameNode) also holds general information, such as game headers.

class chess.pgn.Game

The root node of a game with extra information such as headers and the starting position.

By default the following 7 headers are provided in an ordered dictionary:

>>> game = chess.pgn.Game()
>>> game.headers["Event"]
'?'
>>> game.headers["Site"]
'?'
>>> game.headers["Date"]
'????.??.??'
>>> game.headers["Round"]
'?'
>>> game.headers["White"]
'?'
>>> game.headers["Black"]
'?'
>>> game.headers["Result"]
'*'

Also has all the other properties and methods of GameNode.

headers

A collections.OrderedDict() of game headers.

board()

Gets the starting position of the game as a bitboard.

Unless the SetUp and FEN header tags are set this is the default starting position.

setup(board)

Setup a specific starting position. This sets (or resets) the SetUp and FEN header tags.

class chess.pgn.GameNode
parent

The parent node or None if this is the root node of the game.

move

The move leading to this node or None if this is the root node of the game.

nags = set()

A set of NAGs as integers. NAGs always go behind a move, so the root node of the game can have none.

comment = ''

A comment that goes behind the move leading to this node. Comments that occur before any move are assigned to the root node.

starting_comment = ''

A comment for the start of a variation. Only nodes that actually start a variation (starts_variation()) can have a starting comment. The root node can not have a starting comment.

variations

A list of child nodes.

board()

Gets a bitboard with the position of the node.

It’s a copy, so modifying the board will not alter the game.

san()

Gets the standard algebraic notation of the move leading to this node.

Do not call this on the root node.

root()

Gets the root node, i.e. the game.

end()

Follows the main variation to the end and returns the last node.

starts_variation()

Checks if this node starts a variation (and can thus have a starting comment). The root node does not start a variation and can have no starting comment.

is_main_line()

Checks if the node is in the main line of the game.

is_main_variation()

Checks if this node is the first variation from the point of view of its parent. The root node also is in the main variation.

variation(move)

Gets a child node by move or index.

has_variation(move)

Checks if the given move appears as a variation.

promote_to_main(move)

Promotes the given move to the main variation.

promote(move)

Moves the given variation one up in the list of variations.

demote(move)

Moves the given variation one down in the list of variations.

remove_variation(move)

Removes a variation by move.

add_variation(move, comment='', starting_comment='', nags=())

Creates a child node with the given attributes.

add_main_variation(move, comment='')

Creates a child node with the given attributes and promotes it to the main variation.

Parsing

chess.pgn.read_game(handle, error_handler=<function _raise>)

Reads a game from a file opened in text mode.

By using text mode the parser does not need to handle encodings. It is the callers responsibility to open the file with the correct encoding. According to the specification PGN files should be ASCII. Also UTF-8 is common. So this is usually not a problem.

>>> pgn = open("data/games/kasparov-deep-blue-1997.pgn")
>>> first_game = chess.pgn.read_game(pgn)
>>> second_game = chess.pgn.read_game(pgn)
>>>
>>> first_game.headers["Event"]
'IBM Man-Machine, New York USA'

Use StringIO to parse games from a string.

>>> pgn_string = "1. e4 e5 2. Nf3 *"
>>>
>>> try:
>>>     from StringIO import StringIO # Python 2
>>> except ImportError:
>>>     from io import StringIO # Python 3
>>>
>>> pgn = StringIO(pgn_string)
>>> game = chess.pgn.read_game(pgn)

The end of a game is determined by a completely blank line or the end of the file. (Of course blank lines in comments are possible.)

According to the standard at least the usual 7 header tags are required for a valid game. This parser also handles games without any headers just fine.

The parser is relatively forgiving when it comes to errors. It skips over tokens it can not parse. However it is difficult to handle illegal or ambiguous moves. If such a move is encountered the default behaviour is to stop right in the middle of the game and raise ValueError. If you pass None for error_handler all errors are silently ignored, instead. If you pass a function this function will be called with the error as an argument.

Returns the parsed game or None if the EOF is reached.

chess.pgn.scan_headers(handle)

Scan a PGN file opened in text mode for game offsets and headers.

Yields a tuple for each game. The first element is the offset. The second element is an ordered dictionary of game headers.

Since actually parsing many games from a big file is relatively expensive, this is a better way to look only for specific games and seek and parse them later.

This example scans for the first game with Kasparov as the white player.

>>> pgn = open("mega.pgn")
>>> for offset, headers in chess.pgn.scan_headers(pgn):
...     if "Kasparov" in headers["White"]:
...         kasparov_offset = offset
...         break

Then it can later be seeked an parsed.

>>> pgn.seek(kasparov_offset)
>>> game = chess.pgn.read_game(pgn)

This also works nicely with generators, scanning lazily only when the next offset is required.

>>> white_win_offsets = (offset for offset, headers in chess.pgn.scan_headers(pgn)
...                             if headers["Result"] == "1-0")
>>> first_white_win = next(white_win_offsets)
>>> second_white_win = next(white_win_offsets)

Be careful when seeking a game in the file while more offsets are being generated.

chess.pgn.scan_offsets(handle)

Scan a PGN file opened in text mode for game offsets.

Yields the starting offsets of all the games, so that they can be seeked later. This is just like scan_headers() but more efficient if you do not actually need the header information.

The PGN standard requires each game to start with an Event-tag. So does this scanner.

Writing

If you want to export your game game with all headers, comments and variations you can use:

>>> print(game)
[Event "?"]
[Site "?"]
[Date "????.??.??"]
[Round "?"]
[White "?"]
[Black "?"]
[Result "*"]

1. e4 e5 { Comment } *

Remember that games in files should be separated with extra blank lines.

>>> print(game, file=handle, end="\n\n")

Use exporter objects if you need more control. Exporter objects are used to allow extensible formatting of PGN like data.

class chess.pgn.StringExporter(columns=80)

Allows exporting a game as a string.

The export method of Game also provides options to include or exclude headers, variations or comments. By default everything is included.

>>> exporter = chess.pgn.StringExporter()
>>> game.export(exporter, headers=True, variations=True, comments=True)
>>> pgn_string = str(exporter)

Only columns characters are written per line. If columns is None then the entire movetext will be on a single line. This does not affect header tags and comments.

There will be no newlines at the end of the string.

class chess.pgn.FileExporter(handle, columns=80)

Like a StringExporter, but games are written directly to a text file.

There will always be a blank line after each game. Handling encodings is up to the caller.

>>> new_pgn = open("new.pgn", "w")
>>> exporter = chess.pgn.FileExporter(new_pgn)
>>> game.export(exporter)

NAGs

Numeric anotation glyphs describe moves and positions using standardized codes that are understood by many chess programs. During PGN parsing, annotations like !, ?, !!, etc. are also converted to NAGs.

NAG_NULL = 0
NAG_GOOD_MOVE = 1

A good move. Can also be indicated by ! in PGN notation.

NAG_MISTAKE = 2

A mistake. Can also be indicated by ? in PGN notation.

NAG_BRILLIANT_MOVE = 3

A brilliant move. Can also be indicated by !! in PGN notation.

NAG_BLUNDER = 4

A blunder. Can also be indicated by ?? in PGN notation.

NAG_SPECULATIVE_MOVE = 5

A speculative move. Can also be indicated by !? in PGN notation.

NAG_DUBIOUS_MOVE = 6

A dubious move. Can also be indicated by ?! in PGN notation.

Polyglot opening book reading

chess.polyglot.open_reader(path)

Creates a reader for the file at the given path.

>>> with open_reader("data/opening-books/performance.bin") as reader:
>>>    entries = reader.get_entries_for_position(board)
class chess.polyglot.Entry

An entry from a polyglot opening book.

key

The Zobrist hash of the position.

raw_move

The raw binary representation of the move. Use the move() method to extract a move object from this.

weight

An integer value that can be used as the weight for this entry.

learn

Another integer value that can be used for extra information.

move()

Gets the move (as a Move object).

class chess.polyglot.Reader(handle)

A reader for a polyglot opening book opened in binary mode. The file has to be seekable.

Provides methods to seek entries for specific positions but also ways to efficiently use the opening book like a list.

>>> # Get the number of entries
>>> len(reader)
92954
>>> # Get the nth entry
>>> entry = reader[n]
>>> # Iteration
>>> for entry in reader:
>>>     pass
>>> # Backwards iteration
>>> for entry in reversed(reader):
>>>     pass
seek_entry(offset, whence=0)

Seek an entry by its index.

Translated directly to a low level seek on the binary file. whence is equivalent.

seek_position(position)

Seek the first entry for the given position.

Raises KeyError if there are no entries for the position.

next_raw()

Reads the next raw entry as a tuple.

Raises StopIteration at the EOF.

next()

Reads the next Entry.

Raises StopIteration at the EOF.

get_entries_for_position(position)

Seeks a specific position and yields all entries.

chess.POLYGLOT_RANDOM_ARRAY = [0x9D39247E33776D41, ..., 0xF8D626AAAF278509]

Array of 781 polyglot compatible pseudo random values for Zobrist hashing.

Syzygy endgame tablebase probing

Syzygy tablebases provide WDL (win/draw/loss) and DTZ (distance to zero) information for all endgame positions with up to 6 pieces. Positions with castling rights are not included.

class chess.syzygy.Tablebases(directory=None, load_wdl=True, load_dtz=True)

Manages a collection of tablebase files for probing.

Syzygy tables come in files like KQvKN.rtbw or KRBvK.rtbz, one WDL (.rtbw) and DTZ (.rtbz) file for each material composition.

Directly loads tables from directory. See open_directory.

open_directory(directory, load_wdl=True, load_dtz=True)

Loads tables from a directory.

By default all available tables with the correct file names (e.g. KQvKN.rtbw or KRBvK.rtbz) are loaded.

Returns the number of successfully openened and loaded tablebase files.

probe_wdl(board)

Probes WDL tables for win/draw/loss-information.

Probing is thread-safe when done with different board objects and if board objects are not modified during probing.

Returns None if the position was not found in any of the loaded tables.

Returns 2 if the side to move is winning, 0 if the position is a draw and -2 if the side to move is losing.

Returns 1 in case of a cursed win and -1 in case of a blessed loss. Mate can be forced but the position can be drawn due to the fifty-move rule.

>>> with chess.syzygy.Tablebases("data/syzygy") as tablebases:
...     tablebases.probe_wdl(chess.Board("8/2K5/4B3/3N4/8/8/4k3/8 b - - 0 1"))
...
-2
probe_dtz(board)

Probes DTZ tables for distance to zero information.

Probing is thread-safe when done with different board objects and if board objects are not modified during probing.

Return None if the position was not found in any of the loaded tables. Both DTZ and WDL tables are required in order to probe for DTZ values.

Returns a positive value if the side to move is winning, 0 if the position is a draw and a negative value if the side to move is losing.

A non-zero distance to zero means the number of halfmoves until the next pawn move or capture can be forced, keeping a won position. Minmaxing the DTZ values guarantees winning a won position (and drawing a drawn position), because it makes progress keeping the win in hand. However the lines are not always the most straight forward ways to win. Engines like Stockfish calculate themselves, checking with DTZ, but only play according to DTZ if they can not manage on their own.

>>> with chess.syzygy.Tablebases("data/syzygy") as tablebases:
...     tablebases.probe_dtz(chess.Board("8/2K5/4B3/3N4/8/8/4k3/8 b - - 0 1"))
...
-53
close()

Closes all loaded tables.

UCI engine communication

The Universal Chess Interface is a protocol for communicating with engines.

chess.uci.popen_engine(command, engine_cls=<class 'chess.uci.Engine'>)

Opens a local chess engine process.

No initialization commands are sent, so do not forget to send the mandatory uci command.

>>> engine = chess.uci.popen_engine("/usr/games/stockfish")
>>> engine.uci()
>>> engine.name
'Stockfish 230814 64'
>>> engine.author
'Tord Romstad, Marco Costalba and Joona Kiiski'

The input and input streams will be linebuffered and able both Windows and Unix newlines.

chess.uci.spur_spawn_engine(shell, command, engine_cls=<class 'chess.uci.Engine'>)

Spwans a remote engine using a Spur shell.

>>> import spur
>>> shell = spur.SshShell(hostname="localhost", username="username", password="pw")
>>> engine = chess.uci.spur_spwan_engine(shell, ["/usr/games/stockfish"])
>>> engine.uci()
class chess.uci.Engine(process)
process

The underlying operating system process.

name

The name of the engine. Conforming engines should send this as id name when they receive the initial uci command.

author

The author, as sent via id author. Just like the name.

options

A case insensitive dictionary of Options. The engine should send available options when it receives the initial uci command.

uciok

threading.Event() that will be set as soon as uciok was received. By then name, author and options should be available.

return_code

The return code of the operating system process.

terminated

threading.Event() that will be set as soon as the underyling operating system process is terminated and the return_code is available.

terminate(async=False)

Terminate the engine.

This is not an UCI command. It instead tries to terminate the engine on operating system level, for example by sending SIGTERM on Unix systems. If possible, first try the quit command.

Returns:The return code of the engine process.
kill(async=False)

Kill the engine.

Forcefully kill the engine process, for example by sending SIGKILL.

Returns:The return code of the engine process.
is_alive()

Poll the engine process to check if it is alive.

UCI commands

class chess.uci.Engine(process)
uci(async_callback=None)

Tells the engine to use the UCI interface.

This is mandatory before any other command. A conforming engine will send its name, authors and available options.

Returns:Nothing
debug(on, async_callback=None)

Switch the debug mode on or off.

In debug mode the engine should send additional infos to the GUI to help debugging. This mode should be switched off by default.

Parameters:on – bool
Returns:Nothing
isready(async_callback=None)

Command used to synchronize with the engine.

The engine will respond as soon as it has handled all other queued commands.

Returns:Nothing
setoption(options, async_callback=None)

Set a values for the engines available options.

Parameters:options – A dictionary with option names as keys.
Returns:Nothing
ucinewgame(async_callback=None)

Tell the engine that the next search will be from a different game.

This can be a new game the engine should play or if the engine should analyse a position from a different game. Using this command is recommended but not required.

Returns:Nothing
position(board, async_callback=None)

Set up a given position.

Instead of just the final FEN, the initial FEN and all moves leading up to the position will be sent, so that the engine can detect repetitions.

If the position is from a new game it is recommended to use the ucinewgame command before the position command.

Parameters:board – A chess.Board.
Returns:Nothing
go(searchmoves=None, ponder=False, wtime=None, btime=None, winc=None, binc=None, movestogo=None, depth=None, nodes=None, mate=None, movetime=None, infinite=False, async_callback=None)

Start calculating on the current position.

All parameters are optional, but there should be at least one of depth, nodes, mate, infinite or some time control settings, so that the engine knows how long to calculate.

Parameters:
  • searchmoves – Restrict search to moves in this list.
  • ponder – Bool to enable pondering mode. The engine will not stop pondering in the background until a stop command is received.
  • wtime – Integer of milliseconds white has left on the clock.
  • btime – Integer of milliseconds black has left on the clock.
  • winc – Integer of white Fisher increment.
  • binc – Integer of black Fisher increment.
  • movestogo – Number of moves to the next time control. If this is not set, but wtime or btime are, then it is sudden death.
  • depth – Search depth ply only.
  • nodes – Search so many nodes only.
  • mate – Search for a mate in mate moves.
  • movetime – Integer. Search exactly movetime milliseconds.
  • infinite – Search in the backgorund until a stop command is received.
Returns:

In normal search mode a tuple of two elements. The first is the best move according to the engine. The second is the ponder move. This is the reply expected by the engine. Either of the elements may be None. In infinite search mode or ponder mode there is no result. See stop (or ponderhit) instead.

stop(async_callback=None)

Stop calculating as soon as possible.

Returns:A tuple of the latest best move and the ponder move. See the go command. Results of infinite searches will also be available here.
ponderhit(async_callback=None)

May be sent if the expected ponder move has been played.

The engine should continue searching but should switch from pondering to normal search.

Returns:A tuple of two elements. The first element is the best move

according to the engine. The second is the new ponder move. Either of the elements may be None.

quit(async_callback=None)

Quit the engine as soon as possible.

Returns:The return code of the engine process.

Asynchronous communication

By default all operations are executed synchronously and their result is returned. For example

>>> engine.go(movetime=2000)
BestMove(bestmove=Move.from_uci('e2e4'), ponder=None)

will take about 2000 milliseconds. All UCI commands have an optional async_callback argument. They will then immediately return information about the command and continue.

>>> command = engine.go(movetime=2000, async_callback=True)
>>> command.done()
False
>>> command.result() # Synchronously wait for the command to finish
BestMove(bestmove=Move.from_uci('e2e4'), ponder=None)
>>> command.done()
True

Instead of just passing async_callback=True a callback function may be passed. It will be invoked possibly on a different thread as soon as the command is completed. It takes a Command object as a single argument.

>>> def on_go_finished(command):
...     # Will likely be executed on a different thread.
...     bestmove, ponder = command.result()
...
>>> command = engine.go(movetime=2000, async_callback=on_go_finished)

All commands are queued and executed in FIFO order (regardless if asynchronous or not).

class chess.uci.Command

Information about the state of a command.

done()

Returns whether the command has already been completed.

add_done_callback(fn)

Add a callback function to be notified once the command completes.

The callback function will receive the Command object as a single argument.

The callback might be executed on a different thread. If the command has already been completed it will be invoked immidiately, instead.

result(timeout=None)

Wait for the command to finish and return the result.

A timeout in seconds may be given as a floating point number and TimeoutError is raised if the command does not complete in time.

Info handler

Chess engines may send information about their calculations with the info command. You can register info handlers to be asynchronously notified whenever the engine sends more information. You would usually subclass the InfoHandler class.

class chess.uci.Score

A centipawns or mate score sent by an UCI engine.

cp

Evaluation in centipawns or None.

mate

Mate in x or None. Negative if the engine thinks it is going to be mated.

lowerbound

If the score is not exact but only a lowerbound.

upperbound

If the score is only an upperbound.

class chess.uci.InfoHandler
info

The default implementation stores all received information in this dictionary. To get a consistent snapshot use the object as if it were a threading.Lock().

>>> # Register the handler.
>>> handler = InfoHandler()
>>> engine.info_handlers.append(handler)
>>> # Start thinking.
>>> engine.go(infinite=True)
>>> # Wait a moment, then access a consistent snapshot.
>>> time.sleep(3)
>>> with handler:
...     if "score" in handler.info:
...         print("Score: ", handler.info["score"].cp)
...         print("Mate: ", handler.info["score"].mate)
Score: 34
Mate: None
depth(x)

Received search depth in plies.

seldepth(x)

Received selective search depth in plies.

time(x)

Received new time searched in milliseconds.

nodes(x)

Received number of nodes searched.

pv(moves)

Received the principal variation as a list of moves.

In MultiPV mode this is related to the most recent multipv number sent by the engine.

multipv(num)

Received a new multipv number, starting at 1.

score(cp, mate, lowerbound, upperbound)

Received a new evaluation in centipawns or a mate score.

cp may be None if no score in centipawns is available.

mate may be None if no forced mate has been found. A negative numbers means the engine thinks it will get mated.

lowerbound and upperbound are usually False. If True, the sent score are just a lowerbound or upperbound.

currmove(move)

Received a move the engine is currently thinking about.

currmovenumber(x)

Received a new currmovenumber.

hashfull(x)

Received new information about the hashtable.

The hashtable is x permill full.

nps(x)

Received new nodes per second statistic.

tbhits(x)

Received new information about the number of table base hits.

cpuload(x)

Received new cpuload information in permill.

string(string)

Received a string the engine wants to display.

refutation(move, refuted_by)

Received a new refutation of a move.

refuted_by may be a list of moves representing the mainline of the refutation or None if no refutation has been found.

Engines should only send refutations if the UCI_ShowRefutations option has been enabled.

currline(cpunr, moves)

Received a new snapshot of a line a specific CPU is calculating.

cpunr is an integer representing a specific CPU. moves is a list of moves.

pre_info(line)

Received a new info line about to be processed.

When subclassing remember to call this method of the parent class in order to keep the locking in tact.

post_info()

Processing of a new info line has been finished.

When subclassing remember to call this method of the parent class in order to keep the locking in tact.

pre_bestmove(line)

A new bestmove command is about to be processed.

on_bestmove(bestmove, ponder)

A new bestmove and pondermove have been received.

post_bestmove()

A new bestmove command was processed.

Since this indicates that the current search has been finished the dictionary with the current information will be cleared.

Options

class chess.uci.Option

Information about an available option for an UCI engine.

name

The name of the option.

type

The type of the option.

Officially documented types are check for a boolean value, spin for an integer value between a minimum and a maximum, combo for an enumeration of predefined string values (one of which can be selected), button for an action and string for a textfield.

default

The default value of the option.

There is no need to send a setoption command with the defaut value.

min

The minimum integer value of a spin option.

max

The maximum integer value of a spin option.

var

A list of allows string values for a combo option.

Indices and tables