The following topics are covered:
Natural-based XML parser using dynamic variables.
Functions for
conversion of Natural data structures into DTD definitions;
generation of COMPRESS statements to save a Natural data structure as an XML document;
generation of callback for the Natural-based parser.
The objective of the Natural XML Toolkit is to provide additional XML functionality with Natural and improve the integration of Natural applications with XML.
The Natural XML Toolkit consists of a collection of Natural programs. The Toolkit programs may be integrated into customer applications to provide access to XML data or to deliver data from Natural in XML format.
The Natural XML Toolkit calls the functions listed below:
XML Toolkit Functions
Mapping of Natural Data Definition to DTD and vice versa.
XML Token => NAT
Data After the Natural data structure has been created, the
XML document has to be parsed and saved into the data structure. A Natural
implementation is generated that is capable of saving the given data into the
Natural data structure.
NAT Data => XML Document ("Serialize")
Serialization is the process of taking the data stored in the
Natural data structures and creating an XML document according to the
description provided.
A Natural dialog implements the user interface to the XML Toolkit functions. The DTD will be accessed as a work file and the generated Natural objects will be saved directly to the Natural system file.
This mapping is the first step to bind Natural data structures to XML tags and is required to implement a representation of Natural data as XML tags. The example below shows the mapping as well as some obvious differences between Natural and a DTD.
Press ESC to enter command mode
Mem: EMPL Lib: SYSXTK Type: PARAMETER Bytes: 1072 Line: 0 of: 26
C T Comment
* *** Top of Data Area ***
1 EMPLOYEE
2 ATTRIBUTES_OF_EMPLOYEE
3 PERSONNEL-ID A 8
*
2 FULL-NAME
3 FIRST-NAME A 20
3 NAME A 20
*
2 FULL-ADDRESS
3 C@ADDRESS-LINE I 4
3 ADDRESS-LINE A 20 (1:6)
3 CITY A 20
3 ZIP A 20
3 COUNTRY A 3
*
2 TELEPHONE
3 AREA-CODE A 6
3 PHONE A 15 |
Generated DTD
<!ELEMENT EMPLOYEE (PERSONNEL-ID, FULL-NAME, FULL-ADDRESS, TELEPHONE, INCOME* )>
<!ELEMENT PERSONNEL-ID (#PCDATA ) >
<!ELEMENT FULL-NAME (FIRST-NAME, NAME )>
<!ELEMENT FIRST-NAME (#PCDATA )>
<!ELEMENT NAME (#PCDATA )>
<!ELEMENT FULL-ADDRESS (ADDRESS-LINE*, CITY, ZIP, COUNTRY )>
<!ELEMENT ADDRESS-LINE (#PCDATA )>
<!ELEMENT CITY (#PCDATA )>
<!ELEMENT ZIP (#PCDATA )>
<!ELEMENT COUNTRY (#PCDATA )>
...
The generated DTD will be used later on during serialization to a XML document (see below).
During execution of a Natural program, the content of the data defined in the DEFINE DATA statement will be filled with "real" content. This content will be written to a dynamic variable in XML format during serialization and will use the formerly generated DTD as input.
The XML Toolkit generates the program to serialize the data.
<?xml version="1.0" encoding="iso-8859-1" ?> <EMPLOYEE PERSONNEL-ID="30016509"> <FULL-NAME> <FIRST-NAME>ELSPETH</FIRST-NAME> <NAME>TROWBRIDGE</NAME> </FULL-NAME> <FULL-ADDRESS> <ADDRESS-LINE>91 BACK LANE</ADDRESS-LINE> <ADDRESS-LINE>BILSTON</ADDRESS-LINE> <ADDRESS-LINE>STAFFORDSHIRE</ADDRESS-LINE> <CITY>BILSTON</CITY> <ZIP>ST2 3KA</ZIP> <COUNTRY>UK</COUNTRY> </FULL-ADDRESS> <TELEPHONE> <PHONE>863322</PHONE> <AREA-CODE>0602</AREA-CODE> </TELEPHONE> ...
The mapping of a DTD to Natural data structures again shows differences. The DTD does not specify how many person records will be included in the XML document, therefore the Toolkit assumes that a maximum number of "v" persons will be included. The application programmer might know the exact number and the data structure could be adapted accordingly. A similar limitation exists with the length of the data. The DTD does not include information about the length of the data in a person's record. Therefore the Toolkit creates fields in the data structure with a length of 253, the current maximum.
* DTD E:\SAG\nat\v.r\fnat\SYSXTK\RES\empl.dtd
COMPRESS &1& '<EMPLOYEE'
' PERSONNEL-ID="'EMPLOYEE.PERSONNEL-ID "'
'>' INTO &1& LEAVING NO
/* now the children
COMPRESS &1& '<FULL-NAME'
'>' INTO &1& LEAVING NO
/* now the children
COMPRESS &1& '<FIRST-NAME'
'>'
EMPLOYEE.FIRST-NAME
'</FIRST-NAME>' INTO &1& LEAVING NO
COMPRESS &1& '<NAME'
'>'
EMPLOYEE.NAME
'</NAME>' INTO &1& LEAVING NO
/*
COMPRESS &1& '</FULL-NAME>' INTO &1& LEAVING NO
COMPRESS &1& '<FULL-ADDRESS'
'>' INTO &1& LEAVING NO
/* now the children
FOR &2& = 1 TO EMPLOYEE.C@ADDRESS-LINE
COMPRESS &1& '<ADDRESS-LINE'
'>'
EMPLOYEE.ADDRESS-LINE(&2&)
'</ADDRESS-LINE>' INTO &1& LEAVING NO
END-FOR
...
* DTD E:\SAG\nat\v.r\fnat\SYSXTK\RES\empl.dtd
DECIDE ON FIRST &1&
VALUE 'EMPLOYEE'
RESET INITIAL EMPLOYEE
VALUE 'EMPLOYEE/@PERSONNEL-ID'
/* #REQUIRED
EMPLOYEE.PERSONNEL-ID := &3&
VALUE 'EMPLOYEE/FULL-NAME'
IGNORE
VALUE 'EMPLOYEE/FULL-NAME/FIRST-NAME'
IGNORE
VALUE 'EMPLOYEE/FULL-NAME/FIRST-NAME/$'
EMPLOYEE.FIRST-NAME := &3&
VALUE 'EMPLOYEE/FULL-NAME/NAME'
IGNORE
VALUE 'EMPLOYEE/FULL-NAME/NAME/$'
EMPLOYEE.NAME := &3&
...
The XML Toolkit is another step forward to full XML support with Natural. The XML Toolkit might be extended after the first release. However, the main objective is to implement XML functionality in one of the forthcoming releases as part of Natural's powerful language.
The XML Toolkit only supports fully assembled XML Schema (Layer 1). For detailed information, refer to the W3C recommendation on XML Schema (Layer 1)
When using the XML Toolkit, the following further limitations should be considered.
Data structures which will result in more then approximately 700 data fields and groups will end up with the message:
Input Structure too big
Split up the data structure into smaller sections.
- <include>
Include adds multiple schemas with the same target namespace to a document. The document needs to be included, without any changes.
- <import>
Import adds multiple schemas with different target namespaces to a document. First the document to be imported requires a namespace prefix translation, then the document can be included.
- <redefine>
Redefine selects out specific simple and complex types, groups, and attribute groups from an external schema, and enables you to modify the given specification for your own needs.
Note:
With all of the above elements, only relative URIs are
allowed. Absolute URIs (e.g. http://www.yourdomain.com/your/path or
file://your/path) can not be used.
The external data has to be included into the document. There are no conversions necessary.
If a <![ INCLUDE ] is found, the contained
definition will be used for generation.
If a <![ IGNORE ] is found, the contained
definition will not be used for generation.
The XML Toolkit supports two different kinds of wildcard representations:
Save all subsequent elements or
Save all attributes that are not specified
For wildcard support the following rules and/or limitations apply.
For attributes an attributes_of_<entity-name>
group is generated. All attributes connected to this group are added. The name
of an attribute is saved as a variable name; the content is the content of the
variable.
To add <any> attributes, it is required to add
a variable that contains all attributes not specified.
Note:
The <any> attributes are no "real"
attributes; they are used as a container for the not parsed data and contain
the attribute/value pairs. An <any> attribute is represented
by a ##ANY variable of type (A) dynamic.
Because it can be necessary to access this data, a more specific
name should be used instead of ##ANY followed by a generic number.
It is recommended to add the name of the parent entity and the keyword
ATTR, or ATTRIBUTE. See below for an example:
1 HTML
2 BODY
3 ATTRIBUTES_OF_BODY
4 BGCOLOR (A) DYNAMIC
4 ##ANY_ATTR_BODY (A) DYNAMIC
If, during the parse process, an attribute that is not named
inside the XML Schema is found, the variable name and value will be saved at
the all_attributes_of_<element-name> group as they are, this
means with the standard XML syntax:
<attribute-name1>="<attribute-value1>"
<attribute-name2>="<attribute-value2>" etc.
When serializing, the above string will be added.
To add the <any> data type, it is necessary to
save all subsequent data of an entity, regardless of the names and values of
this entity.
Note:
The <any> entities do not specify "real"
entities; they are used as a container for the not parsed data and contain the
entities with their entire content (attributes, etc.). An
<any> entity is represented by a ##ANY variable
of type (A) dynamic.
Because it can be necessary to access this data, a more specific
name should be used instead of ##ANY followed by a generic number.
It is recommended to add the name of the parent entity. See below for an
example:
1 HTML
2 BODY
3 ATTRIBUTES_OF_BODY
4 BGCOLOR (A) DYNAMIC
4 ##ANY_ATTR_BODY (A) DYNAMIC
3 ##ANY_BODY (A) DYNAMIC
If, during the parse process, an element of type
<any> is found, all subsequent data is collected.
When serializing, all data is taken without changes and is added to the resulting XML document.
Even if the attributes "maxOccurs" and/or
"minOccurs" for xs:any are specified, the Natural
variable implementing xs:any is always a scalar. The Natural
variable may contain data of more then one entity.
The attribute "namespace" for xs:any is
not evaluated, the Natural variable implementing xs:any may
contain entities of different namespaces.
The attribute "processContents" for
xs:any is not evaluated, because the parser used is not
validating.
If a xs:choice or xs:sequence contains
more then one definition of xs:any, the generation ends with an
error, because during parse different <any> containers can
not be recognized.
If a document contains entities that are not specified at the XML
schema, and at the same level xs:any is defined, the Natural
variable implementing xs:any may contain this "nonspecified"
entity data.
During generation of an external data structure, each variable
prefixed with ##ANY will be converted to the specific wildcard
functions:
##ANY_ -> any entity type. Applies to DTDs
and XML Schema.
##ANY_ATTR_ -> any attribute type. Applies to
XML Schema only.
When multiple occurrences (maxOccurs>1) for
unnamed group structures like xs:sequence, xs:choice
or xs:all are defined then the generated source code might need
some manual adaptations. This becomes necessary because unnamed groups usually
do not result in unique xpath names that can be used for parsing.