Entities are also further classified as parsed or unparsed:
A parsed entity contains text, which will be incorporated into the document and parsed if the entity is referenced. A parameter entity can only be a parsed entity.
An unparsed entity contains any kind of data, and a reference to it will result in the application's merely being notified of the entity's presence; the content of the entity will not be parsed, even if it is text. An unparsed entity can only be external.
Internal and external entities
An internal entity has a value that is either a literal string, or a parsed string comprising markup and entities defined in the same document. In contrast, an external entity has a declaration that invokes an external document, thereby necessitating the intervention of an entity manager to resolve the external document reference.
System entities
An entity declaration may have a literal value, or may have some combination of an optional SYSTEMidentifier, which allows SGML parsers to process an entity's string referent as a resource identifier, and an optional PUBLIC identifier, which identifies the entity independent of any particular representation. In XML, a subset of SGML, an entity declaration may not have a PUBLIC identifier without a SYSTEM identifier.
SGML document entity
When an external entity references a complete SGML document, it is known in the calling document as an SGML document entity. An SGML document is a text document with SGML markup defined in an SGML prologue. A complete SGML document comprises not only the document instance itself, but also the prologue and, optionally, the SGML declaration.
Syntax
An entity is defined via an entity declaration in a document's document type definition. For example:
This DTD markup declares the following:
An internal general entity named greeting1 exists and consists of the string Hello world.
An external general entity named greeting2 exists and consists of the text found in the resource identified by the URIfile:///hello.txt.
An internal parameter entity named greeting3 exists and consists of the string ¡Hola!.
An internal general entity named greeting4 exists and consists of the string ¡Hola! means Hello!.
Names for entities must follow the rules for SGML names, and there are limitations on where entities can be referenced. Parameter entities are referenced by placing the entity name between % and ;. Parsed general entities are referenced by placing the entity name between "&" and ";". Unparsed entities are referenced by placing the entity name in the value of an attribute declared as type ENTITY. The general entities from the example above might be referenced in a document as follows: '&greeting1;' is a common test string. The content of hello.txt is: &greeting2; In Spanish, &greeting4;
When parsed, this document would be reported to the downstream application the same as if it has been written as follows, assuming the hello.txt file contains the text Salutations: 'Hello world' is a common test string. The content of hello.txt is: Salutations In Spanish, ¡Hola! means Hello!
A reference to an undeclared entity is an error unless a default entity has been defined. For example:
Additional markup constructs and processor options may affect whether and how entities are processed. For example, a processor may optionally ignore external entities.
Character entities
Standard entity sets for SGML and some of its derivatives have been developed as mnemonic devices, to ease document authoring when there is a need to use characters that are not easily typed or that are not widely supported by legacy character encodings. Each such entity consists of just one character from the Universal Character Set. Although any character can be referenced using a numeric character reference, a character entity reference allows characters to be referenced by name instead of code point. For example, HTML 4 has 252 built-in character entities that do not need to be explicitly declared, while XML has five. XHTML has the same five as XML, but if its DTDs are explicitly used, then it has 253.
