Reporting requirements for XML parsers

Introduction

In this paper I suggest reporting requirements for various categories of XML parser. These requirements are expressed in terms of the XML Infoset.

I consider three common categories of XML parser:

Minimal parser: Reads only the document entity. Does not process any parameter entities. Accepts any well-formed document.
Whole-document-reading (WDR) parser: Reads the whole document, including all parameter and general parsed entities. May reject documents that refer to undeclared entities even when this is technically only a validity error.
Validating parser: A WDR parser which checks validity constraints and satisifies certain additional reporting requirements. May reject invalid documents.

In addition, a parser may support XML Namespaces and/or XML Base.

Properties marked (*) are not explicitly required by the XML 1.0 specification.

Document Information Item
Property	Notes
[children](*)	Only element and PI children are required
[notations]	Only notations used as attribute values or PI targets are required
[entities]	Only unparsed entities used as attribute values are required

Element Information Item
Property	Notes
[local name](*)	May be combined with [prefix]
[prefix](*)	May be combined with [local name]
[children](*)	Only element, PI and unexpanded entity children are required
[attributes]	May be combined with [namespace attributes]
[namespace attributes]	May be combined with [attributes]
[base URI](*)	The URI of the containing entity

Attribute Information Item
Property	Notes
[local name](*)	May be combined with [prefix]
[prefix](*)	May be combined with [local name]
[normalized value]
[attribute type]	Only so that entity and notation values can be identified

Processing Instruction Information Item
Property	Notes
[target]
[content]
[base URI](*)	The URI of the containing entity

Unexpanded Entity Information Item
A WDR parser will only return these if an entity is undeclared or cannot be retrieved. Some WDR parsers reject the document in this case.
Property	Notes
[name]
[entity]

Character Information Item
Property	Notes
[character code]

External Entity Information Item
A WDR parser need only return these if it returns unexpanded entity items that refer to them.
Property	Notes
[name]
[system identifier]
[public identifier]

Unparsed Entity Information Item
Property	Notes
[name]
[system identifier]
[public identifier]
[notation]

Minimal parsers return unexpanded entity items instead of the content of external entities.

Validating parsers must return the following additional property:

Character Information Item
Property	Notes
[element content whitespace]

Parsers supporting XML Namespaces must return the following additional properties:

Element Information Item
Property	Notes
[namespace name](*)
[in-scope namespaces](*)	This may be returned implicitly by means of the [namespace attributes] property

Attribute Information Item
Property	Notes
[namespace name](*)

Furthermore, provided they do provide the [in-scope namespaces] property of elements, they need not return the [namespace attributes] property.

Parsers supporting XML Base must take xml:base attributes into account when computing [base URI] properties.

Richard Tobin, February 2000