index Back to help overviewhome Back to home page

Database formats

Scidb supports the following formats:

Scidb (.sci)

This is Scidb's own database format. Use of this format is strongly recommended because of speed and feature completeness considerations.

Only Scidb's format (and PGN format) is storing games of various chess variants (⇒Supported chess variants). A database is fixed to one chess variant (the Clipbase is an exception).

Furthermore this format is supporting Chess 960 , and Shuffle Chess  positions, in all chess variants.

Scidb databases are always encoded with character set UTF-8. Hence these databases can be used without any complications under all systems and locations.

A database in Scidb's format is divided into several files:

.sci Game data
.scg Game moves, commentaries, annotations, move information, and PGN tags
.scn Player/Event/Site/Annotator data
.scr Overhead for game links
.scj Set of icons

Only the files with suffixes .sci, .scg, and .scn are mandatory, others only if required.

The icon set (.scj) will be created when image files are selected inside the selection dialog for the database icon (⇒ database switcher

Scid (.si4, .si3)

Scid's database format is fully supported because of it's wide adoption. But with this format you cannot use the full capability of Scidb.

annotation Scid databases will be opened only in read-only mode.

In some cases the user has to choose the appropriate encoding when opening the database, to ensure a correct display of names and other text. Newer databases may be UTF-8 encoded, older database may be written with Latin-1 (ISO-8859-1) character set, the Scid format does not provide information about this. But in many cases the automatic detection of the character set should be successful.

This format resembles structurally Scidb's format:

.si4 Game data
.sg4 Game moves, commentaries, annotations, and PGN tags
.sn4 Player/Event/Site/Round data
.ssc Sorting data

note The files with suffixes .sso (Scid Search Option File), .sor (Scid Repertoire File), and .stm (Scid Tree Mask File), do not belong explicitely to a database and will not be regarded by Scidb. Likewise any saved game list (file suffix .txt) will be disregarded.

ChessBase (.cbh, .cbf)

Both formats will be supported; the new format .cbh and the old format .cbf.

The company ChessBase  keeps this format secret, therefore only the reading mode is supported. Not all game data can be read correctly, with some games the message of failed decoding will appear.

annotation ChessBase databases cannot be used as reference database.

Usually this database is encoded with character set CP1252 (or CP850 with old format), but in some cases a manual selection of the the character set when opening the database may be required, to ensure a correct display of names and other text.

A database in ChessBase format is divided into many separate files, the important ones are:

.cbh Game data
.cbg Game moves and variations
.cba Annotations and commentaries
.cbs Source index
.cbp Players index
.cbc Commentators index
.cbt Tournament index
.cbe Team index
.cbj Team information
.cko Opening keys
.cpo Opening keys
.ini Database initialisation file
.htmlHTML document folder

A database in old format consists of only four files:

.cbfGame data and moves
.cbiIndex data
.ckoOpening keys
.cpoOpening keys

PGN (.pgn, .pgn.gz, .zip)

The PGN format (Portable Game Notation) is a common standard for transfering chess games. Opening a PGN file will be done in reading mode, but it is possible to set it writeable (except Zip archives; ⇒ Write protection of databases). With larger PGN game collections it is recommended to convert it into the Scidb format.

According to the PGN standard  a PGN file is encoded with character set Latin-1 (ISO-8859-1). But if a PGN file is starting with an UTF-8 mark (UTF-8-BOM: byte sequence EF BB BF) an UTF-8 encoding will be expected (according to Unicode FAQ ). If the application is detecting an Unicode string when reading the file before a string with a different encoding (except ASCII) was recognized, then also Unicode encoding for the whole file will be expected, but only if not an explicit encoding is chosen when opening the file.

Only the following variants will be accepted if reading FICS games:

wild/2 misc/bishops-only crazyhouse
wild/5 misc/knights-only endings/kbbk
wild/7 misc/little-game endings/kbnk
wild/8 misc/no-queens losers
wild/17 misc/pawns-only odds/*
wild/19 misc/pyramid pawns/little-game
wild/22 misc/queen-rooks pawns/pawns-only
wild/23 misc/queens-only pawns/wild-five
wild/25 misc/rooks-only suicide
wild/26 misc/runaway
wild/fr

A complete specification of this format will be found at Standard: Portable Game Notation Specification and Implementation Guide .

Database archive (.scv)

Scidb's database archive is based on an independent archive format, and contains one (or perhaps several) databases in archived form. This facilitates the shipment of databases through the internet, because it is a single file instead of several ones. This database archive is not limited to a specific database format.

A detailed description of this format is on page Specification of database archive.