A.4 Property Set Definition Requirements (PSDR)

A.4.5 Canonical Grove Representation (CGR)

Subclauses:

Groves are an abstract data representation. Processing systems need not reflect a literal grove in their internal data structures. However, a literal representation of groves is useful for validation, debugging, and, potentially, interchange among different applications. The canonical grove representation (CGR) document type is used to create character representations of groves such that there can be exactly one canonical grove representation for a given grove or hypergrove.

The CGR document type is defined as an SGML document type derived from the HyTime architecture augmented by a set of constraints on how source documents are formed. These constraints ensure that the only differences between CGR documents are differences resulting from differences in node property values. When the constraints are adhered to, CGR documents can be meaningfully compared as character strings.

NOTE 467 The constraints on CGR document sources have also been designed to enable parsing of CGR documents using the standard string parsing features of common programming languages. While CGR documents are SGML documents and can be parsed as such, they can also be parsed as a sequence of delimited records.

A.4.5.1 Canonical grove representation document type

The CGR document type consists of the following element types:

grove: Represents a single grove. Contains a single node element followed by zero or more urefloc elements.
node

CGR documents conform to the reference concrete syntax except that they use a name length of 32.

<!--=================================================================
    Canonical Grove Representation Document Type

    ISO/IEC 10744:1997//DTD Canonical Grove Representation//EN

    =================================================================-->

<?IS10744 ArcBase HyTime>
<!NOTATION HyTime
   PUBLIC "ISO/IEC 10744:1997//NOTATION AFDR ARCBASE
           Hypermedia/Time-based Structuring Language (HyTime)//EN"
>
<!ATTLIST #NOTATION HyTime
   ArcFormA NAME     #FIXED HyTime
   ArcNamrA NAME     #FIXED HyNames
   ArcDocF  NAME     #FIXED HyDoc
   ArcDTD   CDATA    #FIXED "HyTime"
   ArcBridF NAME     #FIXED HyBrid
   ArcAuto  NAME     #FIXED ArcAuto
   ArcOptSA CDATA    #FIXED "locs"
   locs     CDATA    #FIXED "nmsploc"
>
<!NOTATION AFDRMeta
   PUBLIC "ISO/IEC 10744//NOTATION AFDR Meta-DTD Notation//EN"
>
<!ENTITY HyTime
   PUBLIC "ISO/IEC 10744//DTD AFDR Meta-DTD
           Hypermedia/Time-based Structuring Language (HyTime)//EN"
   CDATA AFDRMeta
>

<!ELEMENT
   grove          -- Grove document element --

   (node,urefloc*)
>

<!ELEMENT
   urefloc        -- Unrestricted reference node location address --

   (#PCDATA)      -- Constraint: content of urefloc is "UNKNOWN"
                     when CGR document not generated as part of
                     hypergrove CGR generation process. --
>
<!ATTLIST urefloc
   id       ID       #REQUIRED
   HyTime   NAME     #FIXED nmsploc
   namespc  NAME     #FIXED elements
   locsrc   ENTITY   #REQUIRED
   notname  NAME     #FIXED ignore
>

<!ELEMENT
   node           -- Node in the grove --

   (nodeprop*)
>
<!ATTLIST node
   id             -- Unique ID of node --
      ID          #REQUIRED

   class          -- RCSNM name of the class for this node --
      NAME
      #REQUIRED
>

<!ELEMENT
   nodeprop       -- Node property --

   (#PCDATA|slsep|node)*
                  -- Constraint: Nodes allowed in content only when
                     noderel is "subnode". --
>
<!ATTLIST nodeprop
   rcsnm          -- RCS name of property --
      NAME
      #REQUIRED

   datatype       -- The RCSNM of the datatype --
      NAME     #REQUIRED

   noderel        -- Relationship of nodes in property to node
                     exhibiting property --
      (subnode|urefnode|irefnode|primitive)
      #REQUIRED

   isnull         -- Property is null --
      (isnull)
      #IMPLIED    -- Default: Property is not null --
>

<!ELEMENT
   slsep          -- String list separator --

   - O
   EMPTY
>

<!-- Entities for escaping markup delimiters in data content -->
<!ENTITY lt   "<" -- Left angle bracket -->
<!ENTITY gt   "<" -- Right angle bracket -->
<!ENTITY apos "'" -- Apostrophe (LITA) -->
<!ENTITY amp  "&" -- Ampersand -->

A.4.5.2 Constraints on CGR source construction

CGR documents must conform to the following constraints on their source:

No white space is allowed except as expressly defined in these rules or required by SGML.

Exactly one SPACE character is used to separate parameters that require separation.

All keywords are specified in upper case (only element content and CDATA attribute values may be in mixed case).

The DOCTYPE declaration up to, but not including, the declaration subset open (if any) is specified on a single line as follows:

<!DOCTYPE GROVE PUBLIC "ISO/IEC 10744:1997//DTD Canonical Grove Representation//EN"

The markup declaration close is specified on a line by itself.

If there is a declaration subset (containing entity declarations for other CGR documents), the declaration subset open and declaration subset close delimiters are each specified on lines by themselves. Each entity declaration is specified on a line by itself (see A.4.5.3 Algorithm for assigning IDs to nodes).
Start tags are specified on multiple lines as follows:
1. The STAGO and GI are specified on a single line.
2. Each attribute is specified on a single line by itself. Attributes are specified in the order they occur in the CGR document type attribute list declarations. Fixed attributes are never specified.
3. The TAGC is specified on a line by itself.
For end-tags, the ETAGO and GI are specified on a single line. The TAGC is specified on a line by itself.
Each grove document must consist of exactly one entity (the document entity), except for the external declaration subset, which is always referred to by public identifier as described above.

Primitive property values are represented as character data content of nodeprop elements. The primitive data types are represented by strings as follows:

enum: The RCS name of the enumeration value, in upper case.
string: Character data content exactly as in the property value, with the exception that the characters "<" (left angle bracket), ">" (right angle bracket), "'" (appostrophe), and "&" (ampersand) are replaced by references to the general entities "lt", "gt", "apos", and "amp", respectively. These entities are declared in the CGR docment type.
string list: A list of string values, each value except the last followed by an slsep start-tag.
integer: The string representation of the integer value, base 10.
component name: The RCS name of the component, in upper case.
integer list: Blank-delimited list of integers.
component name list: Blank-delimited lists of component names.

Properties that have no value (because the when clause for the property was not satisfied) exhibit the value "ISNULL" for the isnull attribute.

The IDs for nodes are generated according to the algorithm defined in the next subclause.

A.4.5.3 Algorithm for assigning IDs to nodes

The iref and uref relationship types are represented as SGML ID references in CGR documents. The IDs for nodes must be generated as follows:

Initialize a counter to zero.
For each node element, generate its ID by prepending the string "X" to the string value (base 10) of the counter. Increment the counter by one.
For each urefnode arc to a node, generate that node's ID as for internal nodes. In addition, generate a corresponding urefloc element with that ID. Urefloc elements are listed following the root node in the order they were generated (thus, in the order of first reference to the nodes they represent).

When a CGR document is generated in isolation, it is impossible to acquire the IDs of nodes in other CGR documents for use in the content of urefloc elements. It is also impossible to generate entity names for the CGR documents that would contain those nodes. However, when a CGR document is generated as part of the generation of the canonical grove representation for a complete hypergrove, then the content of urefloc elements must be the IDs assigned to those nodes. In addition, an entity must be declared for each different grove to which unrestricted references are made.

These entities are declared as follows:

Initialize a counter to zero
For each different grove to which an unrestricted reference is made, in the order of first reference, generate an entity name by prepending the string "CGR" to the string value (base 10) of the counter. Increment the counter by one.
For each entity name generated, in the order of generation, create an entity declaration with the following form:
```
<!ENTITY name PUBLIC "-//HYPERCGR//DOCUMENT name//EN" CDATA HyTime
>
```
Where both occurrences of "name" are replaced by the generated entity name.
NOTE 468 There is no requirement, and no effort should be made, to coordinate the public identifiers generated in different grove documents. Thus the same grove document may be referred to by different public identifiers in the different groves that refer to it.

Each entity declaration is included in the internal DOCTYPE declaration subset on a single line, with the declaration close on a line by itself, as shown above.

NOTE 469 It is up to the system that generates a canonical hypergrove to create the mapping catalog that associates the public identifiers generated for grove documents with those documents.

			Copyright © 1992, 1997 International Organization for Standardization. All rights reserved. This electronic document is for use during development and review of International Standards. Official printed copies of International Standards can be purchased from the ISO and the national standards organization of your country.
Next Clause	Previous Clause