XML Tutorial

From Lazarus wiki
Jump to navigationJump to search

Deutsch (de) English (en) español (es) français (fr) magyar (hu) Bahasa Indonesia (id) italiano (it) 日本語 (ja) 한국어 (ko) português (pt) русский (ru) 中文(中国大陆)‎ (zh_CN)

The Extensible Markup Language (XML) is a World Wide Web Consortium (or "W3C") recommended language created to interchange information between different systems.

It is a text based way to store information as opposed to storing the information in a binary format.

Modern data interchange languages such as XHTML, as well as most WebServices technologies, are based on XML. This wiki can only really give a thumbnail overview of XML, with the primary focus being the parsing and use of XML files in Free Pascal applications. If you are interested in a more comprehensive explanation of XML and how it is used, see http://en.wikipedia.org/wiki/XML.

Introduction

Currently there is a set of units that provides support for XML on Free Pascal. These units are called "XMLRead", "XMLWrite" and "DOM" and they are part of the Free Component Library (FCL) from the Free Pascal Compiler. The FCL is already on the default search path for the compiler on Lazarus, so you only need to add the units to your uses clause in order to get XML support. The FCL is not documented currently (October / 2005), so this short tutorial aims at introducing XML access using those units.

The XML DOM (Document Object Model) is a set of standardized objects that provide a similar interface for using XML on different languages and systems. The standard only specifies the methods, properties and other interface parts of the object, leaving the implementation free for different languages. The FCL currently fully supports the XML DOM 2.0 and a subset of XML DOM 3.0 listed here.

Usage Examples

Below there is a list of XML data manipulation examples with growing complexity.

Units in the uses clause

FPC comes with XML units that supports UTF8 and UTF16. They may not have the latest updates, so there is as well a version of those units (prefixed with "laz2_") in the Lazarus package LazUtils. The units are compatible and one can change from to the other just by changing the uses clause.

The units for using the FPC XML support which uses strings encoded in the system encoding are:

  • DOM
  • XMLRead
  • XMLWrite
  • XMLCfg
  • XMLUtils
  • XMLStreaming

The units for using the Lazarus XML support which has full UTF-8 Unicode support are:

  • laz2_DOM
  • laz2_XMLRead
  • laz2_XMLWrite
  • laz2_XMLCfg
  • laz2_XMLUtils
  • laz_XMLStreaming.

Not all of them are needed in every example, though. You will need DOM as it defines several types including TXMLDocument.

Reading a text node

For Delphi Programmers:

Note that when working with TXMLDocument, the text within a Node is considered a separate TEXT Node. As a result, you must access a node's text value as a separate node. Alternatively, the TextContent property may be used to retrieve content of all text nodes beneath the given one, concatenated together.

The ReadXMLFile procedure always creates a new TXMLDocument, so you don't have to create it beforehand. However, be sure to destroy the document by calling Free when you are done.

For instance, consider the following XML:

<?xml version="1.0"?>
<request>
  <request_type>PUT_FILE</request_type>
  <username>123</username>
  <password>abc</password>
</request>

The following code example shows both the correct and the incorrect ways of getting the value of the text node (add the units laz2_XMLRead and laz2_DOM to the used units list):

var
  PassNode: TDOMNode;
  Doc: TXMLDocument;
begin
  try
    // Read in xml file from disk
    ReadXMLFile(Doc, 'test.xml');
    // Retrieve the "password" node
    PassNode := Doc.DocumentElement.FindNode('password');
    // Write out value of the selected node
    WriteLn(PassNode.NodeValue); // will be blank
    // The text of the node is actually a separate child node
    WriteLn(PassNode.FirstChild.NodeValue); // correctly prints "abc"
    // alternatively
    WriteLn(PassNode.TextContent);
  finally
    // finally, free the document
    Doc.Free;
  end;
end;

Note that ReadXMLFile(...) ignores all leading whitespace characters when parsing a document. The section whitespace characters describes how to keep them.

Printing the names of nodes and attributes

If you want to navigate the DOM tree: when you need to access nodes in sequence, it is best to use FirstChild and NextSibling properties (to iterate forward), or LastChild and PreviousSibling (to iterate backward).

For random access it is possible to use ChildNodes or GetElementsByTagName methods, but these will create a TDOMNodeList object which eventually must be freed. This differs from other DOM implementations like MSXML, because the FCL implementation is object-based, not interface-based.

The following example shows how to print the names of nodes to a TMemo placed on a form.

Below is the XML file called 'test.xml':

<?xml version="1.0"?>
<images directory="mydir">
  <imageNode URL="graphic.jpg" title="">
    <Peca DestinoX="0" DestinoY="0">Pecacastelo.jpg1.swf</Peca>
    <Peca DestinoX="0" DestinoY="86">Pecacastelo.jpg2.swf</Peca>
  </imageNode>
</images>

And here the Pascal code to execute the task:

var
  Doc: TXMLDocument;
  Child: TDOMNode;
  j: Integer;
begin
  try
    ReadXMLFile(Doc, 'test.xml');
    Memo.Lines.Clear;
    // using FirstChild and NextSibling properties
    Child := Doc.DocumentElement.FirstChild;
    while Assigned(Child) do
    begin
      Memo.Lines.Add(Child.NodeName + ' ' + Child.Attributes.Item[0].NodeValue);
      // using ChildNodes method
      with Child.ChildNodes do
      try
        for j := 0 to (Count - 1) do
          Memo.Lines.Add(format('%s %s (%s=%s; %s=%s)',
                                [
                                  Item[j].NodeName,
                                  Item[j].FirstChild.NodeValue,
                                  Item[j].Attributes.Item[0].NodeName,  // 1st attribute details
                                  Item[j].Attributes.Item[0].NodeValue,
                                  Item[j].Attributes.Item[1].NodeName,  // 2nd attribute details
                                  Item[j].Attributes.Item[1].NodeValue
                                ]));
      finally
        Free;
      end;
      Child := Child.NextSibling;
    end;
  finally
    Doc.Free;
  end;
end;

This will print:

imageNode graphic.jpg
Peca Pecacastelo.jpg1.swf (DestinoX=0; DestinoY=0)
Peca Pecacastelo.jpg2.swf (DestinoX=0; DestinoY=86)

Populating a TreeView with XML

One common use of XML files is to parse them and show their contents in a tree like format. You can find the TTreeView component on the "Common Controls" tab on Lazarus.

The function below will take a XML document previously loaded from a file or generated on code, and will populate a TreeView with it´s contents. The caption of each node will be the content of the first attribute of each node.

procedure TForm1.XML2Tree(tree: TTreeView; XMLDoc: TXMLDocument);
var
  iNode: TDOMNode;

  procedure ProcessNode(Node: TDOMNode; TreeNode: TTreeNode);
  var
    cNode: TDOMNode;
    s: string;
  begin
    if Node = nil then Exit; // Stops if reached a leaf
    
    // Adds a node to the tree
    if Node.HasAttributes and (Node.Attributes.Length>0) then
      s := Node.Attributes[0].NodeValue
    else
      s := ''; 
    TreeNode := tree.Items.AddChild(TreeNode, s);

    // Goes to the child node
    cNode := Node.FirstChild;

    // Processes all child nodes
    while cNode <> nil do
    begin
      ProcessNode(cNode, TreeNode);
      cNode := cNode.NextSibling;
    end;
  end;
    
begin
  iNode := XMLDoc.DocumentElement.FirstChild;
  while iNode <> nil do
  begin
    ProcessNode(iNode, nil); // Recursive
    iNode := iNode.NextSibling;
  end;
end;

Another example that displays the complete XML structure including all attribute values (note: the long line referencing TreeView has been split so it will word wrap for this wiki; when writing it in code you do not have to break the line unless you like the formatting) :

procedure XML2Tree(XMLDoc:TXMLDocument; TreeView:TTreeView);

  // Local function that outputs all node attributes as a string
  function GetNodeAttributesAsString(pNode: TDOMNode):string;
  var i: integer;
  begin
    Result:='';
    if pNode.HasAttributes then
      for i := 0 to pNode.Attributes.Length -1 do
        with pNode.Attributes[i] do
          Result := Result + format(' %s="%s"', [NodeName, NodeValue]);

    // Remove leading and trailing spaces
    Result:=Trim(Result);
  end;

  // Recursive function to process a node and all its child nodes 

  procedure ParseXML(Node:TDOMNode; TreeNode: TTreeNode);
  begin
    // Exit procedure if no more nodes to process
    if Node = nil then Exit;

    // Add node to TreeView
    TreeNode := TreeView.Items.AddChild(TreeNode, 
                                          Trim(Node.NodeName+' '+ 
                                           GetNodeAttributesAsString(Node)+ 
                                           Node.NodeValue)
                                        );

    // Process all child nodes
    Node := Node.FirstChild;
    while Node <> Nil do
    begin
      ParseXML(Node, TreeNode);
      Node := Node.NextSibling;
    end;
  end;

begin
  TreeView.Items.Clear;
  ParseXML(XMLDoc.DocumentElement,nil);
end;

Modifying a XML document

The first thing to remember is that TDOMDocument is the "handle" to the DOM. You can get an instance of this class by creating one or by loading a XML document.

Nodes on the other hand cannot be created like a normal object. You *must* use the methods provided by TDOMDocument to create them, and later use other methods to put them in the correct place in the tree. This is because nodes must be "owned" by a specific document in DOM.

Below are some common methods from TDOMDocument:

function CreateElement(const tagName: DOMString): TDOMElement; virtual;
function CreateTextNode(const data: DOMString): TDOMText;
function CreateCDATASection(const data: DOMString): TDOMCDATASection; virtual;
function CreateAttribute(const name: DOMString): TDOMAttr; virtual;

CreateElement creates a new element.

CreateTextNode creates a text node.

CreateAttribute creates an attribute node.

CreateCDATASection creates a CDATA section: regular XML markup characters such as <> are not interpreted within the CDATA section. See Wikipedia article on CDATA

A more convenient method to manipulate attributes is to use TDOMElement.SetAttribute method, which is also represented as the default property of TDOMElement:

// these two statements are equivalent
Element.SetAttribute('name', 'value');
Element['name'] := 'value';

And here an example method that will locate the selected item on a TTreeView and then insert a child node to the XML document it represents. The TreeView must be previously filled with the contents of an XML file using the XML2Tree function.

procedure TForm1.actAddChildNode(Sender: TObject);
var
  position: Integer;
  NovoNo: TDomNode;
begin
  {*******************************************************************
  *  Detects the selected element
  *******************************************************************}
  if TreeView1.Selected = nil then Exit;

  if TreeView1.Selected.Level = 0 then
  begin
    position := TreeView1.Selected.Index;

    NovoNo := XMLDoc.CreateElement('item');
    TDOMElement(NovoNo).SetAttribute('nome', 'Item');
    TDOMElement(NovoNo).SetAttribute('arquivo', 'Arquivo');
    with XMLDoc.DocumentElement.ChildNodes do
    begin
      Item[position].AppendChild(NovoNo);
      Free;
    end;

    {*******************************************************************
    *  Updates the TreeView
    *******************************************************************}
    TreeView1.Items.Clear;
    XML2Tree(TreeView1, XMLDoc);
  end
  else if TreeView1.Selected.Level >= 1 then
  begin
    {*******************************************************************
    *  This function only works on the first level of the tree,
    *  but can easily be modified to work for any number of levels
    *******************************************************************}
  end;
end;

Create a TXMLDocument from a string

Given an XML document in string variable MyXmlString, the following code will create it's DOM:

var
  S: TStringStream;
  XML: TXMLDocument;
begin
  S := TStringStream.Create('');
  try
    // Read complete XML document
    ReadXMLFile(XML, S);             
    // Alternatively: read only an XML Fragment
    ReadXMLFragment(AParentNode, S); 
  finally
    S.Free;
  end;
end;

Validating a document

Since March 2007, DTD validation facility has been added to the FCL XML parser. Validation is checking that logical structure of the document conforms to the predefined rules, called Document Type Definition (DTD).

Here is an example of XML document with a DTD:

<?xml version='1.0'?>
<!DOCTYPE root [
<!ELEMENT root (child)+ >
<!ELEMENT child (#PCDATA)>
]>
<root>
  <child>This is a first child.</child>
  <child>And this is the second one.</child>
</root>

This DTD specifies that 'root' element must have one or more 'child' elements, and that 'child' elements may have only character data inside. If parser detects any violations from these rules, it will report them.

Loading such document is slightly more complicated. Let's assume we have XML data in a TStream object:

procedure TMyObject.DOMFromStream(AStream: TStream);
var
  Parser: TDOMParser;
  Src: TXMLInputSource;
  TheDoc: TXMLDocument;
begin
  try
    // create a parser object
    Parser := TDOMParser.Create;
    // and the input source
    Src := TXMLInputSource.Create(AStream);
    // we want validation
    Parser.Options.Validate := True;
    // assign a error handler which will receive notifications
    Parser.OnError := @ErrorHandler;
    // now do the job
    Parser.Parse(Src, TheDoc);
    // ...and cleanup
  finally
    Src.Free;
    Parser.Free;
  end;
end;

procedure TMyObject.ErrorHandler(E: EXMLReadError);
begin
  if E.Severity = esError then  // we are interested in validation errors only
    writeln(E.Message);
end;

Whitespace characters

If you want to preserve leading whitespace characters in node texts, the above method is the way to load your XML document. Leading whitespace characters are ignored by default. That is the reason why the ReadXML(...) function never returns any leading whitespace characters in node texts. Before calling Parser.Parse(Src, TheDoc) insert the line

Parser.Options.PreserveWhitespace := True;

This will force the parser to return all whitespace characters including all the newline characters that exist in an XML document to make it more readable!

The following code is similar to the ReadXMLFile() procedure, but preserves leading whitespace:

procedure ReadXMLFilePreserveWhitespace(out Doc: TXMLDocument; FileName: string);
var
  Parser: TDOMParser;
  Src: TXMLInputSource;
  InFile: TFileStream;
begin
  try
    InFile := TFileStream.Create(FileName, fmOpenRead);
    Src := TXMLInputSource.Create(InFile);
    Parser := TDOMParser.Create;
    Parser.Options.PreserveWhitespace := True;
    Parser.Parse(Src, Doc);
  finally
    Src.Free;
    Parser.Free;
    InFile.Free;
  end;
end;

Note that the xml document Doc is an out parameter created by the local Parser. It is the user's responsibility to free the Doc himself when it is no longer needed.

Streamed reading

DOM-based processing requires the entire document loaded into memory. This may be not desirable, or not possible if document is huge. FCL provides functionality to read XML data one node at a time, using TXMLReader class and its descendants. This is similar to SAX, but works without callbacks. TXMLReader closely resembles .NET XmlReader class. A basic example follows:

uses
  Classes,xmlreader,xmltextreader,xmlutils;

procedure readme(AStream: TStream);
var
  xtr: TXmlReader;
  settings: TXMLReaderSettings;
  inp: TXMLInputSource;
begin
  settings := TXMLReaderSettings.Create;
  try
    settings.PreserveWhiteSpace := True;
    settings.Namespaces := True;
    inp := TXMLInputSource.Create(AStream);
    try
      xtr := TXmlTextReader.Create(inp,settings);
      try
        // Here the reading starts
        while xtr.Read do
        begin
          write(xtr.NodeType:25);
          if xtr.name<>'' then
            write(xtr.Name:9)
          else
            write('*no name* ');
          write(xtr.Value);
          writeln;
          if xtr.NodeType=ntElement then
          begin
            // print attributes
            if xtr.MoveToFirstAttribute then
            begin
              repeat
                writeln('---',xtr.NodeType:21,xtr.Name:10,xtr.Value:10);
              until not xtr.MoveToNextAttribute;
              xtr.MoveToContent;
            end;  
          end;
        end;
      // Cleanup follows  
      finally
        xtr.Free;
      end;
    finally
      inp.Free;
    end;
  finally
    settings.Free;
  end;
end;

Generating a XML file

Below is the complete code to write a XML file. (This was taken from a tutorial in the DeveLazarus blog) Please, remember to include the DOM and XMLWrite units in your uses clause.

unit Unit1;

{$mode objfpc}{$H+}

interface

uses
  Classes, SysUtils, LResources, Forms, Controls, Graphics, Dialogs, StdCtrls,
  DOM, XMLWrite;

type
  { TForm1 }
  TForm1 = class(TForm)
    Button1: TButton;
    Label1: TLabel;
    Label2: TLabel;
    procedure Button1Click(Sender: TObject);
  private
    { private declarations }
  public
    { public declarations }
  end;
  
var
  Form1: TForm1;
  
implementation

{ TForm1 }

procedure TForm1.Button1Click(Sender: TObject);
var
  Doc: TXMLDocument;                                  // variable to document
  RootNode, parentNode, nofilho: TDOMNode;                    // variable to nodes
begin
  try
    // Create a document
    Doc := TXMLDocument.Create;

    // Create a root node
    RootNode := Doc.CreateElement('register');
    Doc.Appendchild(RootNode);                           // save root node
  
    // Create a parent node
    RootNode:= Doc.DocumentElement;
    parentNode := Doc.CreateElement('usuario');
    TDOMElement(parentNode).SetAttribute('id', '001');       // create atributes to parent node
    RootNode.Appendchild(parentNode);                          // save parent node

    // Create a child node
    parentNode := Doc.CreateElement('nome');                // create a child node
    // TDOMElement(parentNode).SetAttribute('sexo', 'M');     // create atributes
    nofilho := Doc.CreateTextNode('Fernando');         // insert a value to node
    parentNode.Appendchild(nofilho);                         // save node
    RootNode.ChildNodes.Item[0].AppendChild(parentNode);       // insert child node in respective parent node
 
    // Create a child node
    parentNode := Doc.CreateElement('idade');               // create a child node
    // TDOMElement(parentNode).SetAttribute('ano', '1976');   // create atributes
    nofilho := Doc.CreateTextNode('32');               // insert a value to node
    parentNode.Appendchild(nofilho);                         // save node
    RootNode.ChildNodes.Item[0].AppendChild(parentNode);       // insert a childnode in respective parent node

    writeXMLFile(Doc, 'test.xml');                     // write to XML
  finally
    Doc.Free;                                          // free memory
  end;
end;

initialization
  {$I unit1.lrs}

end.

The result will be the XML file below:

<?xml version="1.0"?>
<register>
  <usuario id="001">
    <nome>Fernando</nome>
    <idade>32</idade>
  </usuario>
</register>

An example where you don't need to reference an item by index.

procedure TForm1.Button2Click(Sender: TObject);
var
  Doc: TXMLDocument;
  RootNode, ElementNode,ItemNode,TextNode: TDOMNode;
  i: integer;
begin
  try
    // Create a document
    Doc := TXMLDocument.Create;
    // Create a root node
    RootNode := Doc.CreateElement('Root');
    Doc.Appendchild(RootNode);
    RootNode:= Doc.DocumentElement;
    // Create nodes
    for i := 1 to 20 do
    begin
      ElementNode:=Doc.CreateElement('Element');
      TDOMElement(ElementNode).SetAttribute('id', IntToStr(i));

      ItemNode:=Doc.CreateElement('Item1');
      TDOMElement(ItemNode).SetAttribute('Attr1', IntToStr(i));
      TDOMElement(ItemNode).SetAttribute('Attr2', IntToStr(i));
      TextNode:=Doc.CreateTextNode('Item1Value is '+IntToStr(i));
      ItemNode.AppendChild(TextNode);
      ElementNode.AppendChild(ItemNode);

      ItemNode:=Doc.CreateElement('Item2');
      TDOMElement(ItemNode).SetAttribute('Attr1', IntToStr(i));
      TDOMElement(ItemNode).SetAttribute('Attr2', IntToStr(i));
      TextNode:=Doc.CreateTextNode('Item2Value is '+IntToStr(i));
      ItemNode.AppendChild(TextNode);
      ElementNode.AppendChild(ItemNode);

      RootNode.AppendChild(ElementNode);
    end;
    // Save XML
    WriteXMLFile(Doc,'TestXML_v2.xml');
  finally
    Doc.Free;
  end;

Generated XML:

<?xml version="1.0"?>
<Root>
  <Element id="1">
    <Item1 Attr1="1" Attr2="1">Item1Value is 1</Item1>
    <Item2 Attr1="1" Attr2="1">Item2Value is 1</Item2>
  </Element>
  <Element id="2">
    <Item1 Attr1="2" Attr2="2">Item1Value is 2</Item1>
    <Item2 Attr1="2" Attr2="2">Item2Value is 2</Item2>
  </Element>
  <Element id="3">
    <Item1 Attr1="3" Attr2="3">Item1Value is 3</Item1>
    <Item2 Attr1="3" Attr2="3">Item2Value is 3</Item2>
  </Element>
</Root>

Encoding

Starting from FPC version 2.4, the XML reader is able to process data in any encoding by using external decoders. See XML_Decoders for more details.

According to the XML standard, the encoding attribute in the first line of the XML is optional in case the actual encoding is UTF-8 (without BOM - Byte Order Marker) or UTF-16 (UTF-16 BOM).

TXMLDocument has an encoding property since FPC 2.4. It is ignored as WriteXMLFile always uses UTF-8.

  • FPC 2.4 doesn´t generate an encoding attribute in the first line of the XML file
  • FPC 2.6.0 and later explicitly write an UTF8 encoding attribute, as this is needed for some programs that cannot handle the XML without it.

See also

External Links