Modeling JSON Text with YANG
draft-lhotka-yang-json-00

Abstract

This document defines rules for mapping data models expressed in YANG to JSON text. It does so by specifying a procedure for translating the subset of YANG-compatible XML documents to JSON text, and vice versa.

Status of this Memo

This Internet-Draft is submitted in full conformance with the provisions of BCP 78 and BCP 79.

Internet-Drafts are working documents of the Internet Engineering Task Force (IETF). Note that other groups may also distribute working documents as Internet-Drafts. The list of current Internet- Drafts is at http://datatracker.ietf.org/drafts/current/.

Internet-Drafts are draft documents valid for a maximum of six months and may be updated, replaced, or obsoleted by other documents at any time. It is inappropriate to use Internet-Drafts as reference material or to cite them other than as "work in progress."

This Internet-Draft will expire on October 02, 2012.

Copyright Notice

Copyright (c) 2012 IETF Trust and the persons identified as the document authors. All rights reserved.

This document is subject to BCP 78 and the IETF Trust's Legal Provisions Relating to IETF Documents (http://trustee.ietf.org/license-info) in effect on the date of publication of this document. Please review these documents carefully, as they describe your rights and restrictions with respect to this document. Code Components extracted from this document must include Simplified BSD License text as described in Section 4.e of the Trust Legal Provisions and are provided without warranty as described in the Simplified BSD License.

Table of Contents

• 1. Introduction
• 2. Terminology and Notation
• 3. Specification of the Translation Procedure
• 3.1. Names and Namespaces
• 3.2. Mapping XML Elements to JSON Objects
• 3.3. Mapping YANG Datatypes to JSON Values
• 3.3.1. Numeric Types
• 3.3.2. The "string" Type
• 3.3.3. The "boolean" Type
• 3.3.4. The "enumeration" Type
• 3.3.5. The "bits" Type
• 3.3.6. The "binary" Type
• 3.3.7. The "leafref" Type
• 3.3.8. The "identityref" Type
• 3.3.9. The "empty" Type
• 3.3.10. The "union" Type
• 3.3.11. The "instance-identifier" Type
• 3.4. Example
• 3.5. IANA Considerations
• 3.6. Security Considerations
• 3.7. Acknowledgments
• 4. References
• 4.1. Normative References
• 4.2. Informative References
• Author's Address

1. Introduction

The aim of this document is define rules for mapping data models expressed in the YANG data modeling language [RFC6020] to JavaScript Object Notation (JSON) text [RFC4627]. The result can be potentially applied in two different ways: [XML] is used for encoding both protocol messages and configuration data.

1. JSON can be used in the context of the NETCONF protocol [RFC6241] and with existing data models expressed in YANG.
2. Other documents that choose JSON to represent structured data can use YANG for defining the data model, i.e., both syntactic and semantic constraints that the data have to satisfy.

The former use case is not currently possible without further work because all NETCONF and YANG standards so far assume that XML 

JSON mapping rules could be specified in a similar way as the XML mapping rules in [RFC6020]. This would however require solving several problems. To begin with, YANG uses XPath [XPath] quite heavily but XPath is not defined for JSON and such a definition would be far from straightforward.

In order to avoid these technical difficulties, this document employs an alternative approach: it defines a relatively simple procedure which allows to translate the subset of XML that can be modeled using YANG to JSON, and vice versa. Consequently, validation of a JSON text against a data model can done by translating the JSON text to XML, which is then validated according to the rules stated in [RFC6020].

The translation procedure is adapted to YANG specifics and requirements, namely:

1. Translation of YANG namespaces is supported.
2. The information about datatypes of leaf nodes is used for translating the leaf values to the most appropriate JSON datatype.
3. Translation of XML attributes, mixed content, comments and processing instructions is not supported.

Properties [it.ns] and [it.typ] mean that the translation is driven by a YANG data model which must therefore be known in advance.

2. Terminology and Notation

The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this document are to be interpreted as described in [RFC2119].

The following terms are defined in [RFC6020]:

• anyxml
• augment
• container
• data model
• data node
• data tree
• datatype
• identity
• instance identifier
• leaf
• leaf-list
• list
• module
• submodule

The following terms are defined in [XMLNS]:

• local name
• prefixed name
• qualified name

3. Specification of the Translation Procedure

The translation procedure defines a 1-1 correspondence between the subset of YANG-compatible XML documents and JSON text. This means that the translation can be applied in both directions and is always invertible.

The following subsections contain mainly rules for translating XML documents to JSON text. Rules for the inverse translation are stated only where needed, otherwise they can be easily inferred.

XML documents may be translated to JSON text only if they are valid instances of a YANG data model, which must therefore be known beforehand. There is one extra restriction beyond the standard YANG rules: XML documents with mixed content - which is marginal in YANG and allowed only in anyxml nodes - cannot be translated by this procedure.

The semantics of several YANG statements, such as "rpc", "notification", "config", "if-feature" or "default", are more or less specific to NETCONF and may or may not be meaningful in a non-NETCONF context. If such a statement appears in a data model that is used for the translation procedure, then all YANG rules concerning this statement MUST be observed.

For example, an application that uses YANG for validating JSON text may decide to use the "default" statement for defining default values of JSON's primitive datatypes (string, number or boolean). If it does so, the default values SHALL also be taken into account for the validation of semantic constraints such as those defined by "must" statements.

3.1. Names and Namespaces

The local part of a JSON name is always identical to the local name of the corresponding XML element.

Each JSON name lives in a namespace which is uniquely identified by the name of the YANG module where the corresponding data node is defined. If the data node is defined in a submodule, then the namespace identifier is the name of the main module to which the submodule belongs.

Most of the time, the namespace of JSON names is implicit. The namespace MUST be explicitly specified whenever the namespace of a name differs from that of its parent object, or whenever no parent object exists. The namespace MUST NOT be explicitly specified elsewhere.

In YANG terms, the namespace SHALL be specified only in the names of (i) top-level objects and (ii) objects that augment the data tree of another module (see Section 7.15 in [RFC6020]).

Where the namespace has to be specified in JSON text, it SHALL be done in the following form:

        <module name>:<local name>

The translation procedure MUST correctly map YANG namespace URIs to YANG module names and vice versa.

When mapping namespaces from JSON text to XML, the resulting XML document may use default namespace declarations (via the "xmlns" attribute), prefix-based namespace declarations (via attributes beginning with "xmlns:"), or any combination thereof following the rules stated in [XMLNS]. If prefixed names are used, their prefix SHOULD be the one defined by the "prefix" statement in the YANG module where each data node is defined.

3.2. Mapping XML Elements to JSON Objects

XML elements are translated to JSON objects in a straightforward way:

• XML elements that have no siblings of the same name correspond either to a name/value pair or to a JSON object:
  • An XML element which is modeled as a leaf in YANG is translated to a name/value pair and the JSON datatype of the value is derived from the YANG datatype of the leaf (see Section 3.3 for the datatype mapping rules).
  • An XML element which is modeled as a container in YANG is translated to an object.
• A sequence of sibling XML elements with the same name (modeled as a list or leaf-list in YANG) corresponds to a JSON array. If the sequence is modeled as a leaf-list in YANG, then the array elements are primitive values (strings, numbers or booleans) whose type depends on the datatype of the leaf-list (see Section 3.3). If the sequence is modeled as a list in YANG, then the array elements are JSON objects.

3.3. Mapping YANG Datatypes to JSON Values

3.3.1. Numeric Types

YANG numeric types ("int8", "int16", "int32", "int64", "uint8", "uint16", "uint32", "uint64" and "decimal64") are mapped to JSON numbers whose decimal representation is the YANG canonical form of the number. Hexadecimal values MUST be converted to decimal.

3.3.2. The "string" Type

A "string" value is mapped to an identical JSON string, subject to JSON encoding rules.

3.3.3. The "boolean" Type

A "boolean" value is mapped to the corresponding JSON value 'true' or 'false'.

3.3.4. The "enumeration" Type

An "enumeration" value is mapped in the same way as a string except that the permitted values are defined by "enum" statements in YANG.

3.3.5. The "bits" Type

A "bits" value is mapped to a string identical to the lexical representation of this value in XML, i.e., a space-separated list of bit values.

3.3.6. The "binary" Type

A "binary" value is mapped to a JSON string identical to the lexical representation of this value in XML, i.e., base64-encoded binary data.

3.3.7. The "leafref" Type

A "leafref" value is mapped according to the same rules as the type of the leaf being referred to, subject to the same constraints as the XML value.

3.3.8. The "identityref" Type

An "identityref" value is mapped to a string representing the qualified name of the identity. Its namespace MAY be expressed as shown in Figure 1. If the namespace part is not present, the namespace of the name of the JSON object containing the value is assumed.

3.3.9. The "empty" Type

An "empty" value is mapped to '[null]', i.e., an array with the 'null' value being its only element.

This representation was chosen instead of using simply 'null' in order to facilitate the use of "empty" leafs in common programming languages. When used in a boolean context, the '[null]' value, unlike 'null', evaluates to 'true'.

3.3.10. The "union" Type

YANG "union" type represents a choice among multiple alternative types. The actual type of the XML value MUST be determined using the procedure specified in Sec. 9.12 of [RFC6020] and the mapping rules for that type are used.

3.3.11. The "instance-identifier" Type

An "instance-identifier" value is a string representing a simplified XPath specification. It is mapped to an analogical JSON string in which all occurrences of XML namespace prefixes are either removed or replaced with the corresponding module name according to the rules of Section 3.1.

When translating such a value from JSON to XML, all components of the instance-identifier MUST be given appropriate XML namespace prefixes. It is RECOMMENDED that these prefixes be those defined via the "prefix" statement in the corresponding YANG modules.

3.4. Example

Consider a simple data model defined by the following YANG module:

module ex-json {
  namespace "http://example.com/ex-json";
  prefix ej;
  import ietf-inet-types {
    prefix inet;
  }
  container top {
    list address {
      key "seqno";
      leaf seqno {
        type uint8;
      }
      leaf ip {
        type inet:ip-address;
        mandatory true;
      }
    }
    container phases {
      typedef angle {
        type decimal64 {
          fraction-digits 2;
        }
        units "radians";
      }
      leaf max-phase {
        default "6.28";
        type angle;
      }
      leaf-list phase {
        type angle;
        must ". <= ../max-phase";
        min-elements 1;
      }
    }
  }
}

By using the translation procedure defined in this document, we can conclude that the following JSON text is valid according to the data model:

{
  "ex-json:top": {
    "address": [
      {
        "seqno": 1,
        "ip": "192.0.2.1"
      },
      {
        "seqno": 2,
        "ip": "2001:db8:0:1::1"
      }
    ],
    "phases": {
      "phase": [0.79, 1.04, 3.14]
    }
  }
}

Note that the semantic constraint specified by the "must" statement in Figure 2 is satisfied by all elements of the "phase" array because the default value of 6.28 is used for the absent "max-phase" leaf.

3.5. IANA Considerations

TBD.

3.6. Security Considerations

TBD.

3.7. Acknowledgments

The author wishes to thank Martin Bjorklund and Phil Shafer for their helpful comments and suggestions.

4. References

4.1. Normative References

4.2. Informative References

Author's Address