Copyright © 2018 W3C® (MIT, ERCIM, Keio, Beihang). W3C liability, trademark and document use rules apply.
The EXI format is a compact representation that simultaneously optimizes performance and the utilization of computational resources. The EXI format was designed to support XML representation. With a relatively small set of transformations it may also be used for JSON, a popular format for exchange of structured data on the Web.
This section describes the status of this document at the time of its publication. Other documents may supersede this document. A list of current W3C publications and the latest revision of this technical report can be found in the W3C technical reports index at https://www.w3.org/TR/.
This document has been produced by the Efficient Extensible Interchange (EXI) Working Group. The goals of the EXI Format are discussed in the EXI Format document.
Publication as a Working Group Note does not imply endorsement by the W3C Membership. This is a draft document and may be updated, replaced or obsoleted by other documents at any time. It is inappropriate to cite this document as other than work in progress.
Please send comments about this document to the public-exi@w3.org mailing list (Archives).
This document was produced by a group operating under the W3C Patent Policy.
This document is governed by the 1 February 2018 W3C Process Document.
JSON is a popular format for exchange of structured data on the Web and it is specified in [RFC 7159 - The JavaScript Object Notation (JSON) Data Interchange Format] and [ECMA-404 - The JSON Data Interchange Format]. This document describes how the [Efficient XML Interchange (EXI) Format 1.0 (Second Edition)] specification can be used to represent JSON data efficiently in terms of message size and processing.
The EXI for JSON approach is to equivalently convert any well-formed JSON structures to XML event streams (Appendix D Examples shows some examples) that are directly suitable for datatype-aware EXI representation. Lossless round-trip conversion back to the original JSON structures is supported.
The proposed XML event stream results in a compact format — the so-called EXI for JSON (or EXI4JSON) document — that can be read and written with little additional software. That said, Appendix B XML Schema for EXI4JSON provides an XML Schema describing the EXI for JSON document. EXI processors use the schema-informed grammars that stem from this schema.
The EXI Options describe the EXI options that may be used for any EXI document. Negotiation of what options need to be supported by an EXI for JSON implementation are handled externally to the document. This specification makes use of the default options with the following modifications:
EXI Option | Description | Value |
---|---|---|
strict | Strict interpretation of schemas is used to achieve better compactness | true |
schemaId | Identify the schema information, if any, used to encode the body | "exi4json" |
Both EXI Options for strict and schemaId are REQUIRED and cannot be changed. If future versions of EXI for JSON are specified, version identification is reflected in the schemaId value.
Any valid JSON data can be converted to equivalent EXI. Similarly, corresponding EXI streams that conform to the rules and schema of this specification can be converted to equivalent JSON. The following subsections specify how JSON data MUST be represented for equivalent round-trip conversion. This approach is not suitable for arbitrary EXI or XML data.
Prefixes are used throughout this section to designate certain namespaces. The bindings shown below are assumed, however, any prefixes can be used in practice if they are properly bound to the namespaces.
Prefix | Namespace Name |
---|---|
j | http://www.w3.org/2015/EXI/json |
Also, the specification makes use of EXI event terminology and the associated grammar
notation (e.g., SE
stands for Start Element and EE
for End Element) that is fully described in the EXI specification dealing with
EXI Event
Types.
A JSON value is an object
, array
, string
, or
number
, or one of the following three literal names:
true
false
null
.
object
A JSON object
is represented as a j:map
element which
may comprise a set of key/value pairs as it content.
SE(j:map) <!--key/value pairs--> EE
The XML event sequence for a key/value pair is
SE(j:key) <!--value--> EE
Note:
If the key
-name is not valid with respect to
NCName or it conflicts with any existing global element name
in the XML schema (e.g., array
or string
) the
key
part MUST be escaped as subsequently described in
3.1.1 Key-name Escaping.
That said, any escaped character MUST be unescaped to get back the original
JSON key
-name (see D.3 Example 3).
We distinguish two types of escaping:
Conflict with NCName character(s)
Any character CharRef that is not valid in XML names for use within XML names MUST be escaped as follows
CharRef := '_' [0-9]+ '.'
The digits after '_' up to the terminating '.' provide the decimal number of the character's code point.
(e.g., JSON key "1 key
" becomes "_49._32.key
")
Note:
In order to represent '_' itself, the underscore character needs to be written as "_95." (with period character included) for unambiguous parsing.
Conflict with existing EXI4JSON global schema element name
If the
key
-name is one of the reserved EXI4JSON schema element names (map
,array
,string
,number
,boolean
,null
, orother
), then thekey
MUST be prefixed with the following character sequence "_.
"
(e.g., JSON key "map
" becomes "_.map
")
array
A JSON array
is represented as a j:array
element which
may comprise a collection of values.
SE(j:array) <!--values--> EE
string
A JSON string
MAY be represented as a j:string
element.
SE(j:string) CH(string-value) EE
The EXI for JSON transformation rules allow to map a string also to one of the following more-optimized XML event sequences
SE(j:other) SE(j:base64Binary) CH(string-value) EE EE
SE(j:other) SE(j:dateTime) CH(string-value) EE EE
SE(j:other) SE(j:time) CH(string-value) EE EE
SE(j:other) SE(j:date) CH(string-value) EE EE
Note:
The above mentioned choice requires that the string-value is representable by the corresponding EXI datatype.
number
A JSON number
MAY be represented as a j:number
element.
SE(j:number) CH(number-value) EE
The EXI for JSON transformation rules allow to map a number also to one of the following more-optimized XML event sequences
SE(j:other) SE(j:integer) CH(number-value) EE EE
SE(j:other) SE(j:decimal) CH(number-value) EE EE
Note:
The above mentioned choice requires that the number-value is representable by the corresponding EXI datatype.
true
A JSON true
is represented as a j:boolean
element with
the Characters (CH) event content equals "true"
.
SE(j:boolean) CH("true") EE
The following XML schema describes the EXI4JSON document (see also exi4json.xsd).
<?xml version="1.0" encoding="UTF-8"?> <xs:schema xmlns:xs="http://www.w3.org/2001/XMLSchema" elementFormDefault="qualified" targetNamespace="http://www.w3.org/2015/EXI/json" xmlns:j="http://www.w3.org/2015/EXI/json"> <!-- * This is a schema for the XML representation of JSON * * The schema is made available under the terms of the W3C software notice and license * at https://www.w3.org/Consortium/Legal/copyright-software-19980720 * --> <xs:element name="map" type="j:mapType"/> <xs:element name="array" type="j:arrayType"/> <xs:element name="string" type="j:stringType"/> <xs:element name="number" type="j:numberType"/> <xs:element name="boolean" type="j:booleanType"/> <xs:element name="null" type="j:nullType"/> <xs:element name="other" type="j:otherType"/> <xs:complexType name="mapType"> <xs:sequence minOccurs="0" maxOccurs="unbounded"> <!-- any element is a map key which contains the actual value --> <!-- "key": 25 --> <!-- <age> <number>25</number> </age> --> <xs:any processContents="lax" namespace="##targetNamespace"/> </xs:sequence> </xs:complexType> <xs:complexType name="arrayType"> <xs:choice minOccurs="0" maxOccurs="unbounded"> <xs:element ref="j:map"/> <xs:element ref="j:array"/> <xs:element ref="j:string"/> <xs:element ref="j:number"/> <xs:element ref="j:boolean"/> <xs:element ref="j:null"/> <xs:element ref="j:other"/> </xs:choice> </xs:complexType> <xs:simpleType name="stringType"> <xs:restriction base="xs:string"/> </xs:simpleType> <xs:simpleType name="numberType"> <xs:restriction base="xs:double"> <!-- exclude positive and negative infinity, and NaN --> <!-- Note: No real effect for EXI Float datatype --> <xs:minExclusive value="-INF"/> <xs:maxExclusive value="INF"/> </xs:restriction> </xs:simpleType> <xs:simpleType name="booleanType"> <xs:restriction base="xs:boolean"/> </xs:simpleType> <xs:complexType name="nullType"/> <xs:complexType name="otherType"> <xs:choice> <!-- useful types beyond JSON such as binary, date-times, decimal and integer --> <xs:element name="base64Binary"> <xs:simpleType> <xs:restriction base="xs:base64Binary"/> </xs:simpleType> </xs:element> <xs:element name="dateTime"> <xs:simpleType> <xs:restriction base="xs:dateTime"/> </xs:simpleType> </xs:element> <xs:element name="time"> <xs:simpleType> <xs:restriction base="xs:time"/> </xs:simpleType> </xs:element> <xs:element name="date"> <xs:simpleType> <xs:restriction base="xs:date"/> </xs:simpleType> </xs:element> <xs:element name="integer"> <xs:simpleType> <xs:restriction base="xs:integer"/> </xs:simpleType> </xs:element> <xs:element name="decimal"> <xs:simpleType> <xs:restriction base="xs:decimal"/> </xs:simpleType> </xs:element> </xs:choice> </xs:complexType> </xs:schema>
This section discusses a number of key decision points. A rationale for each decision is given and background information is provided.
other
Datatype RepresentationsCompared to the basic JSON datatypes and the corresponding EXI datatype mapping
(i.e., exi:string, exi:double, and
exi:boolean) the element other
allows for other EXI
datatype representations: namely exi:base64Binary, exi:dateTime,
exi:dateTime, exi:time,
exi:date, exi:integer,
and exi:decimal.
The selection of these additional datatypes is based on their foreseen efficiency and potential usage in JSON documents.
JSON text may be encoded in UTF-8, UTF-16, or UTF-32 (see JSON Character Encoding). EXI for JSON matches the JSON specification in that it does not provide an explicit label for the included characters.
If possible without loss of correctness, processors are recommended to use the default UTF-8 for maximum interoperability when creating JSON documents.
EXI for JSON defines a set of predefined EXI Options beyond the default EXI Options.
The default EXI value for strict is false to permit event items not declared in the schemas.
The main reason to set strict to true in the EXI for JSON context is to reduce specification and code complexity while at the same time allowing for simple implementations. In section 3. Representing JSON data using EXI it is specified how to map an EXI4JSON stream to JSON. Allowing strict to be false would require to deal with unexpected elements and/or attributes and would make the specification more complex while at the same time increase code complexity. The working group concluded that strict being false does not provide any benefit in this context.
Besides that strict being true increases compactness and allows for realizing more-optimized processors with less code.
The schemaId is used to identify the schema information used for processing the EXI stream. The value "exi4json" has been chosen to identify the schema in Appendix B XML Schema for EXI4JSON.
The EXI for JSON structure has been changed compared to the previous publication. A JSON key is not represented anymore as attribute. Instead it is transformed to an element with the JSON value as nested element.
The reason for this change is to allow for dedicated XML schema definitions (which were not possible before). This change implied escaping (see 3.1.1 Key-name Escaping) with the positive side effect to generate valid XML streams.
Note:
EXI streams would not need escaping.
This example illustrates a simple JSON document with a numbered value and an array of strings.
JSON | EXI4JSON | |
---|---|---|
{ "keyNumber": 123, "keyArrayStrings": [ "s1", "s2" ] } | ↔ |
<j:map xmlns:j="http://www.w3.org/2015/EXI/json"> <j:keyNumber> <j:number>123</j:number> </j:keyNumber> <j:keyArrayStrings> <j:array> <j:string>s1</j:string> <j:string>s2</j:string> </j:array> </j:keyArrayStrings> </j:map> |
This is example illustrates nested JSON values such as objects and arrays.
JSON | EXI4JSON | |
---|---|---|
{ "glossary": { "title": "example glossary", "GlossDiv": { "title": "S", "GlossList": { "GlossEntry": { "ID": "SGML", "SortAs": "SGML", "GlossTerm": "Standard Generalized Markup Language", "Acronym": "SGML", "Abbrev": "ISO 8879:1986", "GlossDef": { "para": "A meta-markup language, used to create markup languages such as DocBook.", "GlossSeeAlso": [ "GML", "XML" ] }, "GlossSee": "markup" } } } } } | ↔ |
<j:map xmlns:j="http://www.w3.org/2015/EXI/json"> <j:glossary> <j:map> <j:title> <j:string>example glossary</j:string> </j:title> <j:GlossDiv> <j:map> <j:title> <j:string>S</j:string> </j:title> <j:GlossList> <j:map> <j:GlossEntry> <j:map> <j:ID> <j:string>SGML</j:string> </j:ID> <j:SortAs> <j:string>SGML</j:string> </j:SortAs> <j:GlossTerm> <j:string>Standard Generalized Markup Language</j:string> </j:GlossTerm> <j:Acronym> <j:string>SGML</j:string> </j:Acronym> <j:Abbrev> <j:string>ISO 8879:1986</j:string> </j:Abbrev> <j:GlossDef> <j:map> <j:para> <j:string>A meta-markup language, used to create markup languages such as DocBook.</j:string> </j:para> <j:GlossSeeAlso> <j:array> <j:string>GML</j:string> <j:string>XML</j:string> </j:array> </j:GlossSeeAlso> </j:map> </j:GlossDef> <j:GlossSee> <j:string>markup</j:string> </j:GlossSee> </j:map> </j:GlossEntry> </j:map> </j:GlossList> </j:map> </j:GlossDiv> </j:map> </j:glossary> </j:map> |
This is an example with a keyname "a number"
which is not valid with
respect to NCName and uses 3.1.1 Key-name Escaping.
JSON | EXI4JSON | |
---|---|---|
{ "a number": 1 } | ↔ |
<j:map xmlns:j="http://www.w3.org/2015/EXI/json"> <j:a_32.number> <j:number>1</j:number> </j:a_32.number> </j:map> |
This section describes best practices in how to map knowledge of JSON data into dedicated XML schema definitions. Dedicated XML schema definitions (compared to the pre-defined "exi4json" schemaId specified in this document) allow for an even more efficient representation by keeping compatibility on a structural basis.
Note: JSON schema is used in the examples but the practices are not limited to JSON schema language.
Editorial note | |
Add information about JSON schema "additionalProperties": false/true |
A JSON object is mapped to an element
with complexType and the name attribute set to map
. The child elements appear in all
so that they can appear in any order. The attribute minOccurs
is set to "0" if a child element is not required.
JSON schema | XML schema definition | |
---|---|---|
{ "type": "object", "properties": { "firstname": { "type": "string" }, "lastname": { "type": "string" }, "age": { "type": "number" } }, "required": [ "lastname", "age" ] } | → |
<element name="map"> <complexType> <all> <element minOccurs="0" name="firstname"> <complexType> <sequence> <!--unnamed-format--> <element name="string" type="string"/> </sequence> </complexType> </element> <element name="age"> <complexType> <sequence> <element name="number" type="double"/> </sequence> </complexType> </element> <element name="lastname"> <complexType> <sequence> <!--unnamed-format--> <element name="string" type="string"/> </sequence> </complexType> </element> </all> </complexType> </element> |
A JSON array is mapped to an element
with complexType and the name attribute set to array
. The child elements appear in a sequence
and the attribute minOccurs
is set to "0" and maxOccurs
to "unbounded".
The child elements itself depend on the items type.
JSON schema | XML schema definition | |
---|---|---|
{ "type": "array", "items": { "type": "number" } } | → |
<element name="array"> <complexType> <sequence maxOccurs="unbounded" minOccurs="0"> <element name="number" type="double"/> </sequence> </complexType> </element> |
A JSON string is mapped to an element with the name attribute set to string
and typed as string.
JSON schema | XML schema definition | |
---|---|---|
{ "type": "string" } | → |
<element name="string" type="string"/> |
A JSON number is mapped to an element with the name attribute set to number
and typed as double.
Editorial note | |
Differentiate between "number" and "integer". |
JSON schema | XML schema definition | |
---|---|---|
{ "type": "number" } | → |
<element name="number" type="double"/> |
A JSON boolean is mapped to an element with the name attribute set to boolean
and typed as boolean.
JSON schema | XML schema definition | |
---|---|---|
{ "type": "boolean" } | → |
<element name="boolean" type="boolean"/> |
A JSON null is mapped to an element with the name attribute set to null
and typed as empty complexType.
JSON schema | XML schema definition | |
---|---|---|
{ "type": "null" } | → |
<element name="null"> <complexType/> </element> |
A JSON enumeration is mapped to an element based on the desired type.
JSON schema | XML schema definition | |
---|---|---|
{ "type": "string", "enum": ["red", "amber", "green"] } | → |
<element name="string"> <simpleType> <restriction base="string"> <enumeration value="red"/> <enumeration value="green"/> <enumeration value="amber"/> </restriction> </simpleType> </element> |
This document is the work of the Efficient XML Interchange (EXI) WG.
Members of the Working Group are (at the time of writing, sorted alphabetically by last name):
Carine Bournez, W3C/ERCIM (staff contact)
Don Brutzman, Web3D Consortium
Michael Cokus, MITRE Corporation
Fernández, Javier D., Vienna University of Economics and Business
Joerg Heuer, Siemens AG
Sebastian Käbisch, Siemens AG
Takuki Kamiya, Fujitsu Laboratories of America (chair)
Richard Kuntschke, Siemens AG
Matsukura, Ryuichi, Fujitsu Laboratories
Don McGregor, Web3D Consortium
Daniel Peintner, Siemens AG
Liam Quin, W3C/MIT
Mohamed Zergaoui, INNOVIMAX
The EXI Working Group would like to acknowledge the following former members or external experts for their leadership, guidance and expertise they provided throughout the process of creating this document (sorted alphabetically by last name):
Yusuke Doi
Youenn Fablet
Bruce Hill
Rumen Kyusakov
Stephen Williams
Brett Zamir
The EXI Working Group especially acknowledges and thanks Liam Quin (W3C/MIT) for his guidance and insight throughout the development of the EXI family of recommendations.