Dynamic XML: DOM and SAX

1
Dynamic XMLDOM and SAX

• John Campbell
• White Star Software

Tony Lavinio PSC Labs
Since Version 2
Presentation bp0580
2
Introduction

• Two XML Models
• DOM ? Document Object Model
• SAX ? Simple API for XML
• Differences in Objective
• Differences in Process

3
Johns Background

• 17 Years working with Progress
• (since he was 30)
• Instructor, Author, Programmer
• Moderate XML Knowledge

4
Tonys Background

• 16 years Progress Background
• (coding since he was 12that was young back then
  ?)
• Second time working for PSC
• Author of Many Progress tools (Dictionary,
  Results, you name it) and some Vertical
  Applications

5
Objective

• To create dynamic, flexible XML
• document reading and writing algorithms
• for distinct source and target structures.
• To compare the two models for handling XML

6
Objective Detail

• Read an incoming document with one structure and
  covert it to a known DB structure (DOM and SAX)
• Create an outgoing document from a DB structure
  in the recipients format (DOM)

7
DOM Overview

• Document for populating a DB

8
The Progress DOM

• A Widget Hierarchy
• Begin with the document
• Start at the root
• Work to the leaf nodes
• Examples
• DB to table to field to data (dynamic buffers)
• Window to frame to fields (widgets)

9
DOM Syntax

• Widget types
• x-document
• The container node
• x-noderef
• For pointing to all nodes within the document
• Always pointed to by handles

10
Dynamic Widget Syntax

• Begin with Create of the object
• Use methods to
• Navigate
• Use attributes to
• Query the node
• Populate
• Extract information

11
Creating DOM Objects
define variable hDocument as handle. define
variable hRoot as handle. create
x-document hDocument. create x-noderef hRoot.

• Create the document
• Create a node for the root

12
Methods

• Primary ones for normal documents
• Several others exist

load(ltdocument-namegt) get-document-element(lthandl
egt) get-child(lthandlegt,ltintegergt) get-attribute(
ltcharactergt)
13
Top-level Method

• load(lttypegt,
• ltdocument-namegt ltaddressgt,
• ltlogical-vargt)
• Loads an XML file, and creates a document
  hierarchy
• Type file or memptr
• Name character, address mempointer
• Logical validate the document or not

14
Retrieving the First Node

• get-document-element(lthandlegt)
• Retrieves the root element of the document

15
Loading a Document
define variable hDocument as handle. define
variable hRoot as handle. define variable
OK as log. create x-document
hDocument. create x-noderef hRoot. ok
hDocumentload("file",XMLFile,false). hDocumentge
t-document-element(hRoot).

• This loads the document but does not validate it
• Then gets the root node

16
Other Primary Navigation Methods

• get-child(lthandlegt, ltintegergt)
• Gets the nth child of the current node
• Some nodes have only one child
• Most have many children some have none

17
Getting the Attributes

• get-attribute(ltcharactergt)
• A node may have multiple attributes
• Each name can be derived from the node
• See following slides

18
Attributes of XML Objects

• num-children
• Number of child nodes to this node
• attribute-names
• Names of all the attributes for this node
• name
• Name of a current node
• node-value
• Value of a current node

19
DB Model

• Designed to map from incoming XML doc to DB
• As well as from DB to external document model

20
DB Structure
Field-Name Label Type Format
Purpose -------------- ------ ---- -------
--------------- ObjectType Type char
x(25) Type of Object NameIn Name
char x(25) Incoming Name ObjectName
Object char x(25) Outgoing Name Parent
Parent char x(25) Parent Active
Active log yes/no Is it active?
21
Sample of Data Content

• Type ObjectName
  NameIn
• --------------------- ---------------
  --------------
• Table Customer
  Client
• Field City
  ClientCity
• Field CustNum
  CustNo
• KeyField CustNum
  Customer
• XMLSchema Customer
• XMLTable Client
  Customer
• XMLField ClientCity
  City
• XMLField CustNo
  Custnum
• XMLQuery for each customer
  no-lock
• XMLSField State
  ProvState
• XMLSfield Name
  Name
• XMLSfield Address
  addressline

22
Objective

• Translate an incoming document into Progress data
• Convert Progress data to an outgoing document

23
Principles

• Read the schema for the mapping from the root
  (the document name) to the table to the fields
• This example is the inbound document

24
Supporting Code
define variable hTTSchema as handle
no-undo. define variable hDBTable as handle
no-undo. define variable hDBField as handle
no-undo. define variable TableName as char
no-undo. create temp-table hTTSchema. find first
mapping no-lock where ObjectType "xmlschema"
and active. TableName ObjectName. create buffer
hDBTable for table TableName. / e.g. customer
/ for each mapping no-lock where ObjectType
"XMLSField" / schema field / and parent
TableName / point to the DB field in the base
record / hDBField hDBTablebuffer-field(Objec
tName). / add the temp table field like the
base field / hTTSchemaadd-like-field(hDBField
name,hDBField). end. hTTSchematemp-table-prepare(
TableName). / set the schema / run xmlparser
(input-output table-handle HTTSchema,"Address","te
xt.xml").
25
The Parser

• The main component is an internal procedure which
  is recursively called
• It gets the name of the node, checks it to see if
  it is a field
• If it gets a flag indicating its a new record,
  a DB record is created
• It then populates the field with the incoming data

26
The Code

• define input parameter phNode as handle.
• define variable nodepointer as integer.
• define variable hCurrentNode as handle.
• define variable hText as handle.
• define variable NodeName as char.
• define variable AttrCounter as int.
• create x-noderef hCurrentNode.
• create x-noderef hText.

27
Looping Through the Doc
repeat nodepointer 1 to phNodenum-children
phNodeget-child(hCurrentNode,nodepointer).
/ get the name to see if it is a field in
the T-T / if hCurrentNodeattribute-names
ltgt "" then do AttrCounter 1 to
num-entries(hCurrentNodeattribute-names)
NodeName entry(AttrCounter,
hCurrentNodeattribute-names).
/ is it really a field ? / if
valid-handle(hTTBuffer) then hTTField
httBufferbuffer-field(NodeName) no-error.
28
Prepare to Populate the DB Field

• if valid-handle(hTTField) then
• hTTFIeldbuffer-value
• hCurrentNodeget-attribute(
• entry(AttrCounter,hCurrentNodeAttribute-names))
  .
• / next, see if it is time to create a new record
  /
• NodeName hCurrentNodename
• if NodeName CreateNodeName / from mapping
  table /
• then hTTBufferBuffer-create().
• else if valid-handle(hTTBuffer) then
• / check to see if we have a field to populate /
• hTTField hTTBufferBuffer-field(NodeName)
  no-error.

29
Populate the DB Field

• if valid-handle(hTTField) and NodeName
  hTTFIeldname
• then do
• / get the text value of the current node and
  put it in the field /
• hCurrentNameget-child(hText,1).
• hTTFieldbuffer-value hTextnode-value.
• end.
• / go through all the children of the current
  node /
• run GetLevels (hCurrentNode).
• end. / repeat /
• end procedure.

30
Some Notes

• The value of this model is the ability to move
  back and forth in the document if need be.
• Documents

31
A Sample Incoming Doc

• lt?xml version "1.0" encoding "UTF-8"?gt
• ltMessageSet xmlnsxsi"http//www.w3.org/2001/XMLS
  chema-instance" xsinoNamespaceSchemaLocation"mes
  sage.xsd" Version"1.0"gt
• ltMessagegt
• ltMessageCreategt
• ltDategt06262002lt/Dategt
• ltTimegt113818lt/Timegt
• ltSource Participant "D"gt
• ltDlrCodegt9261lt/DlrCodegt
• lt/Sourcegt
• ltTarget Participant "F"gt
• ltMgmtCodegtTALlt/MgmtCodegt
• lt/Targetgt
• lt/MessageCreategt
• ltNetwork NtwrkID "FSRV"gt
• ltDategt06032002lt/Dategt
• ltTimegt113819lt/Timegt
• lt/Networkgt

32
The Data to be stored

• Note names used by mapping table

ltMessageTypegt ltRequest SourceID
"N0626113897098" MessageID "20993"gt ltBusProcess
gt ltAddModAddressgt ltAddress AddressCode"1"gt
ltNamegtCirque du Soleillt/Namegt ltAd
dressLinegt455 Cote de Neigeslt/AddressLinegt ltAddr
essLinegtBasementlt/AddressLinegt ltAddressLinegtRoom
1lt/AddressLinegt ltCitygtQuebeclt/Citygt ltProvStat
egtQClt/ProvStategt ltCountrygtCANlt/Countrygt ltPosta
lZipgtH8J3I4lt/PostalZipgt lt/Addressgt ltTaxCodegtQC
lt/TaxCodegt lt/AddModAddressgt lt/BusProcessgt lt/Re
questgt lt/MessageTypegt lt/Messagegt
33
DOM Summary

• Using the Progress DOM, it is possible to parse
  incoming files and create outgoing files with
  relative ease.
• You must know dynamic widgets rather well.
• You need to understand the DOM structure
  reasonably well.
• You need know only a modest amount about XML.

34
DOM Plusses and Minuses

• Always have full context available
• Can go backwards and forwards in document
• Can read and write through DOM
• Uses large amount of memory
• May be overkill for just loading

35
SAX (Simple API for XML)

• SAX implemented as callbacks.
• Start the parser, and it calls internal
  procedures as needed as each node is read.
• Pick the data out of the stream as it flows by.
  Once its passed, its gone.
• Dont call us, well call you model.

36
SAX Plusses and Minuses

• Very thrifty for memory
• Callback interface makes for modular code
• Must maintain own context
• Cant go backwards
• Current implementation read-only

37
SAX Strategy

• When we see a ltCustomergt node, well start a new
  customer record.
• Otherwise, if we see an ltelementgt that has the
  same name as a field, well put the content of
  that element into the like-named field.

38
DOM Customer XML example

• define temp-table Custt like Customer.
• create x-document hDoc.
• create x-noderef hRoot.
• create x-noderef hTable.
• create x-noderef hField.
• create x-noderef hText.
• hBuf buffer Custthandle.
• hDocload("file", "cust.xml", false).
• hDocget-document-element(hRoot).

39
DOM Customer XML cont

• repeat nodepointer 1 to hRootnum-children
• hRootget-child(hTable,nodepointer).
• if hTablename ltgt "Customer" then next.
• create Custt.
• do counter 1 to hTablenum-children
• hTableget-child(hField,counter).
• if hFieldnum-children ltgt 1 then next.
• hDBField hBufbuffer-field(hFieldname)
  no-error.
• hFieldget-child(hText,1).
• if hTextname ltgt "text" then next.
• if not error-statuserror then
• hDBFieldbuffer-value hTextnode-value.
• end.
• end.

40
SAX Customer XML Example

• define temp-table Custt like Customer.
• create sax-reader hParser.
• / the callbacks are in this same file /
• hParserhandler this-procedure.
• hBuf buffer Custthandle.
• hParserset-input-source("file", "cust.xml").
• hParsersax-parse().

41
SAX startElement Callback

• procedure startElement
• define input parameter namespaceURI as
  character.
• define input parameter localName as
  character.
• define input parameter qName as
  character.
• define input parameter attributes as handle.
• if localName "Customer" then create Custt.
• else do
• hField hBufbuffer-field(localName) no-error.
• if error-statuserror then hField ?.
• end.
• end procedure.

42
SAX characters Callback

• procedure characters
• define input parameter charData as memptr.
• define input parameter numChars as integer.
• if valid-handle(hField) then
• hFieldbuffer-value get-string(charData,1).
• end procedure.

43
SAX endElement Callback

• procedure endElement
• define input parameter namespaceURI as
  character.
• define input parameter localName as
  character.
• define input parameter qname as
  character.
• / If we dont do this, white space between
• elements will be picked up /
• hField ?.
• end procedure.

44
Loading a Document

• With SAX, you dont load a document, you aim the
  parser at it and let it rip.
• create sax-reader hParser.
• hParserhandler this-procedure.
• hParserset-input-source("file", "cust.xml").
• hParsersax-parse().
• You can also parse from the incoming stream in
  WebSpeed, or from a MEMPTR.

45
Which to Use?

• It depends
• DOM for navigation.
• SAX for scalability.

46
Further Resources

• Progress External Program Interfaces
• Chapter 11, XML Support
• Chapter 12, Simple API for XML
• Progressions, published by White Star Software
• xml_at_peg.com email group

47
Side Benefits of Using XML

• You dont have to worry about internationalization
  .
• Convenient way of packaging schema and data
  together.
• Lots of good tools available, such asSonic
  Softwares Stylus Studio

48
Questions?
49
(No Transcript)