<\body> From the high-level point of view, it is often useful to manipulate database values as other objects than raw strings. In , an additional layer is added to the basic database API which allows you to specify encoding and decoding schemes between raw strings and user defined data formats for specific types of entries and attributes. More precisely, the extension introduces the following tables: <\description> >For each ``kind'' of database ( for bibliographies), specify a finite list of supported entry types ( , , for bibliographies). >For each entry type, specify the format of the fields: mandatory fields, alternative fields and/or optional fields. See for examples. >For each triple of an attribute, an entry type and an encoding scheme, specify a keywords which specifies how to encode/decode field values. The two possible values implemented in are and . , > For the keyword values in , specify the actual routines to be used for encoding and decoding. <\explain> <|explain> Specify the encoding/decoding scheme to be used inside . This value is used as the third component of triples in . Implemented encoding/decoding schemes are (no encoding/decoding) and . <\description> >Specifies the entry type. <\explain> <|explain> The default behaviour is customized so as to encode/decode field values according to the current encoding/decoding scheme.